Eisl §uu.§i.§§rz§ 2521:523 :. r. .__.
* ‘35....2 iv... .f :1... 1; .

. Dirt. a. .,{.1D._FI.X..£7KT€

ix. .a. 5...» 9. a.

. .
\.t\.l- L ‘n
3b.: .r.§.§: . . _ . : ,
31.“; :1: , . . .
. . . . . , irridah...
_ . *fivaC.
‘If’.vI-U.‘pfl

LT.

LEI?

.. 2.
5...: .
21.173. :7:
.. .53.}: E: ..
5:1...
.lii- - Pu. .
'4 ‘5.» ,3
5.35).."
21.5... .
...:.!.f.¢

. 1.1
.3! .7.
7:- o

11 .
. 2.”... n.“flmEn~.I.
£425! .1: . 1;. fig
Eiratauui= Eng...‘

.1. 5.51:. #1
.1...

1.9. .523
a .r
.,
A... ..
2:, ..
55?}...42
.\l.i.kl 1.1.1:“

.2327]. {I
«9.xflk.£.fl:¥u
).{)i.n..7{.i.:
9:53 4
5:5».
31...}:

:zu..r,.
5.1.3..

3...... 7c:
7" .1
x .

. 3.. .J. c .
. . . . p}. \«1 .u. 4:
. , , $.21: .9»; .13 .r 5.:

2.“ aria) ..}a.!:.¢tl.5‘lx

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

“:8 lIllllllllllllllllll

This is to certify that the

dissertation entitled

Evaluating Residential Probation

For Drug Involved Felony Offenders
presentedby

Phyllis Ann Zold

has been accepted towards fulfillment
of the requirements for

Ph.D. degreein Social Science

% Wiwi

/ / mjor professor

 

Date 12/2/1999

MSU it an Afflrman'w‘ Anion/Equal Oppvrlum'ry Inslimnon 0712771

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

til

i’

Muir i 0 9004'

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRC/DateDue.p65-p.15

EVALUATING RESIDENTIAL PROBATION
FOR DRUG INVOLVED FELONY OFFENDERS

BY

Phyllis Ann Zold

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
School of Criminal Justice

1 999

ABSTRACT

EVALUATING RESIDENTIAL PROBATION
FOR DRUG INVOLVED FELONY OFFENDERS

By

Phyllis Ann Zold

Partially in response to the pressures caused by prison overcrowding
the use of probation has expanded. Over the past ten years the probation
population has increased almost 300%. Not only is the pOpulation growing
in size the type of offender on probation has also changed consisting of
offenders convicted of felonies and a greater proportion are also drug
dependent. To manage the changing probation population various models
of intermediate sanctions have been deveIOped. Intensive Supervision
Probation programs (ISP) are one example. Several studies have been
conducted in efforts to determine the effectiveness of ISP programs. In
general, research shows that intensive monitoring alone is not effective in
producing positive outcomes. Instead, successful programs combine
substance abuse treatment with surveillance.

This study examines the impact of a uniquely structured residential
ISP program designed to provide specific levels of both intensive
monitoring and substance abuse treatment for drug involved felony

offenders. Several outcome measures are considered: treatment access;

length of time in treatment; subsequent drug use; and recidivism.

The study is designed as a prospective observational or cohort study
and uses existing data from several sources for conducting secondary data
analysis. Several statistical techniques are utilized. Cox and logistic
regression are used for testing the complex interrelationships among the
data elements. For examining less complex relationships, univariant and
bivariant tools are applied.

This research demonstrates that the residential probation program
examined is more effective in ensuring that drug involved felony offenders
comply with court mandated treatment compared to other types of
probation programs. Moreover, offenders sentenced to residential
probation are somewhat more likely to remain drug and alcohol free during
the first year of follow—up. Overall, time in treatment is proven to be a
significant predictor of all outcome measures. Data also suggests that
improved outcomes can be achieved among these probationers with minor
modifications in the residential probation’s program policies. This research
also supports the claim that to improve outcomes ISP programs need to
change the way in which they are intensive by shifting the emphasis away
from incapacitation and punishment toward a more integrated approach of
intervention and substance abuse treatment. The “intensity" of supervision

needs to focus on intensifying the monitoring of treatment compliance.

 

 

DEDICATED IN LOVING MEMORY
to my father

Abraham Zold

>>> who always believed in me <<<<
and
>>> to whom I apologize -- for not finishing just a little sooner <<<

“SHAL OM’

iv

ACKNOWLEDGMENTS

I would like to thank a number of people who without their support
this dissertation would not have been completed. First and foremost, I
thank my husband, Earl Kilboum for his tireless patience and personal
sacrifices throughout the past decade of this process, which are too
numerous to mention. Dr. Albert Price, my master thesis advisor, I thank for
both “insisting” that I continue on in graduate school after completing a
masters program and for initiating contact with the residential probation
program which provided the ideas and structure in this dissertation.

I am very much indebted to my committee chair, Dr. Merry Morash
for her calming, committed and outstanding guidance and for stepping in
during the eleventh hour, when after three separate dissertation attempts,
all seemed to be lost. I thank each of my committee members for their
individual and special contributions. Dr. Maxwell, I thank for her advice on
methodological issues and equally as important, for her perpetual
encouragement during the frantic moments when the end seemed no where
in sight. Dr. Hudzik, Iwish to thank for initiating the “Chutzpah” necessary
to complete this process. Dr. Broman, I thank for “hanging in there” for me
when it really mattered.

I am particularly grateful to Professor Thomas Tucker at the

University of Kentucky, for his statistical expertise, guidance and
“impromptu tutoring” in helping me to understand and resolve many
statistical issues. I especially thank him for his kind patience while
enduring all of my many questions.

Equally important to this process are those who allowed me access
to the data that was necessary to conduct this dissertation. To protect the
anonymity of the probation and treatment programs, unfortunately these
individuals cannot be named. However, I thank the directors and staff of
the Department of Adult Probation, the Department of Community
Corrections and the residential probation program of the county in which
this study was conducted, for providing access to the probation files. In
addition, I thank the director and staff of the substance abuse treatment
program for assuming the tedious and time-consuming task of collecting
and furnishing the treatment data.

Without the enormous support I received in the work environment,
the writing of this document would not have been possible. I am
appreciative of my supervisor, Dr. Denise Oleske, for her advocacy and
insightful advice in preparing this document. I am very much indebted to
my colleagues, Cheryl Allen-Toole and Jennifer Laskowski, for their
heartfelt encouragement and for doing all that they could to keep me

inspire during the arduous writing process. Thanks ladies!

I am also grateful to my children, Kelli Dufty, Heather Kretzschmer
and Dustin Kilboum, for their encouragement, sacrifices, and who each
contributed to this accomplishment in their own special way. To my
parents, Abraham and Shirley Zold, I thank you for all that you provided me
throughout my life which made this possible. Mom, there are no words to
appropriately express my sincere gratitude for your perpetual
encouragement to complete this process and for the emotional support you
provided when times were especially tough in Dad’s absence. Dad, I am
lovingly grateful for the precious memories which gave me the desire and

motivation to finish, and to your memory I dedicate this dissertation.

vii

TABLE OF CONTENTS

LIST OF TABLES ............................................ x

LIST OF FIGURES ......................................... xiii

I. INTRODUCTION ........................................... 1

Overview of Issues: ..................................... 1

Study Objectives and Scope: .............................. 6

II. LITERATURE REVIEW .................................... 10
Background and Characteristics of

Intensive Supervision Probation: .......................... 10

Effectiveness of ISP Programs: ...................... 13

Compulsory Treatment: ................................. 22

Effectiveness of Compulsory Treatment: ............... 28

The Addiction Process and Recovery: ...................... 33

The Need for Program Evaluations and Change: .............. 38

III. THE RESEARCH SETTING ................................ 42

Overview: ............................................ 42

Environmental Context: ................................. 43

The Intervention Program: ............................... 47

IV. METHODS ............................................. 52

Introduction: .......................................... 52

Research Questions and Hypotheses: ...................... 53

Research Design and Study Population: .................... 58

The Sample: .......................................... 60

Variables and Measures: ................................ 64

Method of Analysis: .................................... 76

V. RESULTS AND FINDINGS ................................. 81

Descriptive Analysis: ................................... 82

Offender Characteristics at the Time of Sentencing: ...... 82

Sociodemographic Characteristics ............... 83

viii

Criminal Justice and Substance

Abuse Characteristics ......................... 85

Criminal Justice Characteristics: Post Sentencing: ....... 88

Substance Abuse Characteristics: Post Sentencing: ...... 99

Descriptive Summary: ............................ 104

Testing the Hypotheses: ................................ 105
Recidivism: ..................................... 1 10

Substance Abuse Treatment Access and Retention: ..... 122

Alcohol and Drug Relapse: ......................... 130

Summary: ........................................... 143

VI. CONCLUSION ........................................ 146
Overview: ........................................... 146
Limitations: .......................................... 148
Discussion of Findings and Policy Recommendations: ........ 149
Treatment Access ................................ 149

Length of Time in Treatment ....................... 153

Substance Abuse Relapse ......................... 157

Recidivism ..................................... 163

Direction for Future Research: ........................... 167

VII. BIBLIOGRAPHY ....................................... 173
General References .................................. 187

ix

Table 1

Table 2

Table 3

Table 4

Table 5

Table 6
Table 7

Table 8

Table 9

Table 10a

Table 10b

Table 1 1a

Table 1 1 b

LIST OF TABLES

Population and Demographics of the Research County . . . . 44
Sociodemographic Characteristics ................... 84

Criminal Justice and Substance Abuse Characteristics
at Sentencing .................................... 86

Criminal Justice Characteristics:
Pre and Post Sentencing .......................... 90

Weeks Surviving Additional Court Sanctions:

Technical Violations and New Felony Crimes .......... 94
Weeks Surviving Technical Violations ................ 96
Weeks Surviving New Felony Crimes .............. .99

Substance Abuse Characteristics:
Pre and Post Sentencing ........................ 101

Weeks in Substance Abuse Treatment ............... 104

Length of Time Until Court Sanctioned Failure:
Full Model .................................... 112

Length of Time Until Court Sanctioned Failure:
Reduced Model ................................ 114

Length of Time Until Court Sanctioned
Technical Violations: Full Model .................... 116

Length of Time Until Court Sanctioned
Technical Violations: Reduced Model .............. 117

 

Table 12a

Table 12b

Table 13a

Table 13b

Table 14

Table 15

Table 16a

Table 16b

Table 17

Table 18a

Table 18b

Length of Time Until New Felony Crimes:
FuII Model ....................................
Length of Time Until New Felony Crimes:

Reduced Model

Offenders Receiving Substance Abuse Treatment:
Full Model ....................................
Offenders Receiving Substance Abuse Treatment:

Reduced Model

Weeks In Substance Abuse Treatment:
ANOVA Results

Weeks Surviving Alcohol and THC Relapse

Length of Time Until Alcohol and THC Relapse:
FuII Model

Length of Time Until Alcohol or THC Relapse:
Reduced Model

Weeks Surviving Drug Relapse Other Than
Alcohol and THC

Length of Time Until Drug Relapse:
Full Model ....................................
Length of Time Until Drug Relapse:

Reduced Model

xi

 

120

121

.125

126

.130

134

135

136

................................ 140

Figure 1

Figure 2
Figure 3
Figure 4
Figure 5

Figure 6

LIST OF FIGURES

Weeks Surviving Addition Court Sanctions:

Technical Violations and New Felony Crime ........... . 93
Weeks Surviving Technical Violations ................ 95
Weeks Surviving New Felony Crime .................. 98
Weeks in Substance Treatment .................... 103
Weeks Surviving Alcohol and THC Relapse ........... 133

Weeks Surviving Drug Relapse
Other Than Alcohol or THC ....................... 139

xii

I. INTRODUCTION

Overview of Issues:

The greatest rise in incarceration in the United State’s history has
taken place in the past three decades, contributing to significant prison
overcrowding (Lurigio, 1997). In the past decade alone, incarceration rates
have more than doubled, increasing 150% from 1980 to 1991 (Bureau of
Justice Statistics, 1992a) and reaching an astounding rate of 645 per
100,000 in 1997, up from 458 per 100,000 in 1990 (Bureau of Justice
Statistics, 1998). By mid-year1997, 1 in every 155 US. residents was
incarcerated in federal, state or local facilities (Bureau of Justice Statistics,
1998)

This increase has in part been due to sweeping revisions in
sentencing legislation and a proliferation of stricter laws which led not only
to a higher percentage of offenders being sentenced to prison, but to their
being sentenced with longer terms and less probability for release on
parole as well (Irwin and Austin, 1987). Despite prison capacity increases
of more than 50% in the past decade (Bureau of Justice Statistics, 1992a),
by early 1990, correctional systems in forty (40) states were operating
under court order or consent decrees to reduce prison overcrowding

(Maguire and Flanagan, 1991). With this nationwide problem, the need for

 

less expensive and flexible alternatives has never been greater.

Partially in response to the pressures caused by the growing prison
population, the use of probation has expanded (Geerken and Hayes, 1993;
Cunnif and Shilton, 1991; Lurigio, 1997). The probation population
increased 5 to 7% each year from 1985 to 1990 (Petersilia and Turner,
1993a) and has continued by an average of 3% each year since 1990
(Bureau of Justice Statistics, 1997b). Over the past ten (10) years the
population has increased almost 300% (Petersilia, 1997). By 1990, 2.5
million adults were on probation in the United States, representing two-
thirds (2/3), of all persons under correctional supervision (Camp and Camp,
1993; Petersilia, 1993a). At year end 1996, over three (3) million adults
were under state or federal probation (Bureau of Justice Statistics, 1997b).
The probation population is now so large that the US. Department of
Justice estimates that nearly 2 % of all US. adult citizens are under
probation supervision on any one day (Bureau of Justice Statistics, 1996a).

Not only is the probation population growing in size, the type of
offender on probation has also changed. Probation is no longer a
sentencing alternative reserved for the first-time misdemeanants or petty
offenders (Cunnif and Shilton, 1991). More of the current probation
population consists of offenders convicted of felonies than misdemeanors

(Petersilia and Turner, 1993a). For example, it is estimated that in 1995 the

probation population consisted of 1.5 million felons as compared to only
one million misdemeanants (Bureau of Justice Statistics, 1997b).
Moreover, of the current probation population a greater proportion are drug
dependent (Petersilia and Turner, 1992; Guynes, 1988; Cunniff and
Shilton, 1991). It is estimated that over half of the probation population are
drug-involved either as users, sellers or both (Langan and Cunniff, 1992).

With recidivism rates among felony probationers as high as 65% in
some jurisdictions, the use of probation as an alternative for these
offenders has been called into question and has become controversial in
recent years (Geerken and Hayes, 1993). According to Geerken and
Hayes (1993), those who are critical of probation’s effectiveness argue that
a significant percentage of all felony offenders diverted from incarceration
continues to commit crimes, is responsible for a significant amount of all
reported crime, and rarely reform. However, standard probation is not
structured or intended to manage this type of offender (Petersilia and
Turner, 1993a), nor is it adequately funded to do so. Adding to the problem,
despite the unprecedented growth and changes in the probation population,
probation budgets have not grown (Langan, 1994). “Nationally, probation
receives less than 10% of state and local government corrections funding,
even though probation supervises two out of every three correctional

clients” (Petersilia, 1995).

Critics of probation may not be justified. The fact is probation is now
being used for the more serious offender. Standard probation cannot and
should not be expected to be capable of responding to the current
demands. It was never structured to do so. It could be argued that
probation itself is not inadequate, but rather the structure and
implementation of probation are. According to Judge Burton Roberts,
Administrative Judge of the Bronx Supreme and Criminal Courts, “Nothing
is wrong with probation. It is the execution of probation that is wrong”
(Klein, 1997). While probation programs are often criticized as inadequate
and an ineffective part of corrections (Clear and Hardyman, 1990), the
concept of probation remains attractive. Probation does have many
advantages over incarceration, e.g., lower cost, increased opportunities for
rehabilitation, reduced risk of criminal socialization, and the possibility for
offenders to continue working (Lurigio, 1997; Petersilia, 1997).

In many states, the increased proportion of more serious offenders
among the probation population, prison crowding, and the need for more
flexible and cost effective sentencing alternatives generated interest in
experimenting with other sentencing options that offered more intense
levels of supervision (Clear and Hardyman, 1990; Thompson, 1990;
Petersilia and Turner, 1993a). During the past decade many models of

“intermediate sanctions,” such as house arrest, electronic monitoring and

intensive supervision were developed. Intensive Supervision Probation
(ISP) programs, often referred to as intensive supervision, alternative
sanctions, or intermediate supervision, (Petersilia and Turner, 1993b;
Petersilia, 1997) were designed in direct response to prison crowding
(Fulton and Gendreau, 1995; Clear and Hardyman, 1990; Byme; 1990).
Like other models of sentencing alternatives, ISP programs were
designed to be community-based sanctions that were tougher than
probation but less severe and expensive than prison (Tonry and Lynch,
1996). These models assume that a significant number of felons sent to
prison are too serious for probation but could be handled in a community
setting and that both prison diversion and probation enhancement are cost
effective. Moreover, ISP programs are believed to offer social and
altruistic benefits; e.g., keeping families together, allowing offenders to
continue working and avoiding harmful effects of prison (Lurigio, 1997).
This new wave of intermediate sanction had great appeal that captured the
attention of both conservative and liberal policy makers. For
conservatives, it provided an opportunity to “get tough” by increasing the
control over offenders without adding costs to corrections. For liberals, it
provided a potential strategy for diverting offenders from prison without

appearing to be “soft on crime” (Byme, 1990).

 

 

Study Objectives and Scope:

Several studies have been conducted in efforts to determine the
effectiveness of various ISP program models and identify program
characteristics which best predict improved client outcomes. (Discussion of
select studies follows in the literature review.) Although findings are many,
research shows that the level of monitoring alone is not effective in
producing positive outcomes (Langan, 1994; Petersilia,1993). Instead,
successful programs combine substance abuse treatment with surveillance
and target an appropriate offender subgroup (Petersilia and Turner,
1993a). This makes sense when considering that drug-involved offenders
constitute a disproportionate share of repeat offenders and will continue to
pose a substantial threat to society if not treated (U.S. Cong. House, 1991
p.76). Without treatment a significant proportion of drug-involved offenders
will continue drug-seeking behavior and become reinvolved in criminal
activity within just a few months (Wexler et al., 1988). Unfortunately, only a
small proportion of these offenders actually receive treatment (Bureau of
Justice Statistics, 1998). Moreover, there is overwhelming evidence that
suggests the “length of time” in substance abuse treatment is the single
best predictor of positive outcomes (Simpson and Sells, 1982; Barr and
Antes, 1981; Holland, 1983; DeLeon, 1984; DeLeon 1988) and further,

compulsory treatment is an effective means in getting offenders into

treatment to achieve positive outcomes (Maddux, 1988; Hubbard, et al.,
1988; McGlothIin, et al., 1977b; 1978).

What remains unknown and continues to be questioned is: what
levels of both substance abuse treatment and surveillance are required to
maximize outcomes, reduce drug-abuse behavior and ultimately
recidivism? Will different combinations of treatment and surveillance
produce different results in different populations? By modifying and
retesting various program models, the chances for improving the
effectiveness of ISP programs will increase. With the growing use of
probation as a sentencing alternative coupled with the increasing
proportion of drug-involved felony offenders, there are reasons and a need
to continue experimenting with community-based sanctions and evaluate
the effectiveness of different models.

This program evaluation study is a response to this need. The study
was designed to examine in some detail the impact of a uniquely structured
residential ISP program on treatment access, substance abuse relapse and
recidivism and compares this model to a variety of other models. The
residential probation program selected for this outcome evaluation is of
particular interest because the program was designed to provide specific
levels of both intensive monitoring and substance abuse treatment for all

drug-involved offenders. Thus, the program incorporates the tenents of

 

deterrence theory with strategies to address the key issues identified in the
literature as barriers to successful ISP programs. By providing on-slte
substance abuse treatment, the residential probation program removes the
barrier of limited treatment capacity and provides a structure to ensure the
integration of treatment and punishment.

Several outcome measures are considered in this study: treatment
access; length of time in treatment; subsequent drug use; and recidivism“.
The analysis to examine recidivism was affected by the reduced sample
size for which data was available. This presented limitations in drawing
meaningful conclusions from the measures of recidivism. However,
because of the known relationship between drug abuse and crime and the
role of treatment in the reduction of crime, measurements of “treatment
access,” “length of time in treatment” and “subsequent drug use” should be
considered as important and good indicators of subsequent criminal
activity.

The importance of this evaluation is twofold. First, as called for in
the probation literature, it provides the ability to test the effectiveness of a
different combination of treatment and surveillance in reducing drug-abuse

behavior. Second, if different combinations of treatment and surveillance

 

1

A complete discussion of the outcome variables can be found in Chapter IV, “ Methods”
of this document.

 

can be shown to reduce drug use among offenders then this information
can then be used by local policy makers to refine current ISP models and
practices in efforts to improve the effectiveness of these programs. Better
models for measuring the effects of these programs on recidivism can then

be developed and tested further.

II. LITERATURE REVIEW

Background and Characteristics of Intensive Supervision Probation:

The development of ISP programs'was based on the propositions of
deterrence theory. Deterrence theory postulates that punishment or the
threat of punishment will inhibit undesirable behavior of those disposed to
commit crime in the future. According to Wright (1991) the success of
deterrence is based on either the severity of punishment or the
enforcement of punishment for the subsequent offences. The impact of
enforcement depends on the probability of being detected and the certainty
of getting caught and punished must be high before individuals will be
deterred from criminal behavior. Consistent with deterrence theory, the
basic premise of intensive supervision is that increased surveillance will act
as a constraint on probationers and the likelihood of detection will act as a
deterrent to crime (Langan, 1994).

Building on deterrence theory, ISP programs are developed based
on four propositions (Petersilia and Turner, 1990). First, ISP programs
propose that intensive monitoring not only makes it difficult for the offender
to commit new crimes, but it also discovers technical violations. Second,
technical violations are worth monitoring in that they signal an offender is

“going bad” and are proxy measures for future criminal behavior. Third, if

10

 

an offender’s probation is revoked for a technical violation, then the system
may be preventing future crime. Finally, if offenders know that their
behavior is closely monitored and probation is revoked for violations, then
they will want to avoid violations.

Acclaimed by many as the most promising criminal justice innovation
in decades, advocates believe that ISP programs have the potential to
alleviate prison crowding at less cost than increasing prison capacity
without compromising public safety (Byme, Lurigio, and Baird, 1989; Fulton
and Gendreau, 1995). According to Petersilia and Turner (1992), ISP
programs are not only better able than prison or probation to fit the
punishment with the crime, but they also may be the “best hope” for
relieving prison crowding and ensuring public safety. Thus, support for ISP
programs has grown and by 1990 every state had developed some form of
ISP (US. General Accounting Office, 1990).

Despite their popularity, there is no standard or generic ISP program
(Turner et al., 1992) and no two programs are exactly alike (Byme, 1986;
Fulton and Gendreau, 1995). Various jurisdictions may emphasize
different goals, differ with respect to phiIOSOphy and methodologies, utilize
different supervision strategies and staffing patterns, serve disparate target
populations and differ in their ability to implement the program as planned

(Byme, Lurigio, and Baird, 1989; Byme, 1990). The acronym alone says

11

little about any one program’s characteristics except that the supervision
provided by the program is more intense than standard probation.
Moreover, there is no agreement on how many supervisory contacts are
necessary for a probation program to be considered intensive (Petersilia
and Turner 1992). “If there is one common element in the array of ISP
programs, it is the emphasis on surveillance and control of offenders . . . ”
(Byme, 1990). Petersilia and Turner (1993b) echo this observation noting,
the only common characteristic of ISP programs is that they involve more
supervision making them more “intensive” (Petersilia and Turner, 1993b).
Thompson (1987) describes the “intensive” in intensive probation
supervision as having five other dimensions beyond frequency of contacts.
In addition to probation officers have multiple contacts with probationers,
supervision is focused. The monitoring activities are aimed at regulating
specific behaviors, for example, drug use, travel and employment. Second,
supervision is ubiquitous. Probation officers usually conduct random drug
tests and unannounced home visits. Third, supervision is graduated
whereby offenders are monitored more closely at the beginning of the
probationary period and monitoring activities become less frequent over
time. Fourth, supervision is strictly enforced and any noncompliance with
probation conditions is swiftly and severely penalized. And finally,

supervision is coordinated by officers who are part of an autonomous unit

12

of probation and are specially trained.

Most programs therefore include some combination of multiple
weekly contacts with a supervising officer with strict enforcement of
probation conditions, place offenders on smaller case loads, require they
participate In work, submit to random urine and alcohol testing, perform
community service and attend treatment (Turner et al, 1992). Moreover,
programs do tend to share common goals as well; to provide an alternative
to incarceration, reduce prison crowding and incarceration costs, to punish
and rehabilitate the offender, and to enhance probation services (Austin et

al., 1990; Fulton and Gendreau, 1995).

Effectiveness of ISP Programs:

Although they are described as a recent innovation, ISP projects
were initiated with Law Enforcement Assistance Administrative (LEAA)
funds in the mid-1960's and early 1970's . These programs were largely
probation-management tools designed to determine the most effective
caseload size needed to achieve rehabilitation (Lurigio, 1997; Banks and
Rardln, 1977). They were not designed to divert offenders from prison but
rather were directed toward offenders who normally would have received
regular probation. The focus of these earlier programs was on

reintegrating the offender into the community (Lurigio, 1997). The

13

 

 

programs were based on two main assumptions: more intensive
supervision will result in more successful outcomes; and the number of
cases in a caseload determines the intensity of supervision received by a
single case (Clear and Hardyman; 1990; Banks and Rardin, 1977). Follow-
up studies of the earliest ISP programs documented disappointing results:
offenders in smaller caseloads did not receive higher levels of supervision;
offenders who were more intensely monitored were not more successful
than offenders receiving less intense monitoring; and finally, the strict
regulations of the programs had both positive and negative effects on
adjustment to supervision (Clear and Hardyman, 1990). In contrast, ISP
programs developed in the 80's and continued through the 90's,
emphasized surveillance and control. These programs either targeted
offenders who were classified as too serious for regular probation but not
necessarily appropriate for incarceration; or offenders who, without the ISP
option, would be sentenced to prison (Bureau of Justice Assistance, 1988).
The most widely implement and highly publicized programs
developed in the 1980's were the Georgia and New Jersey projects, each
with different and distinct models (Lurigio, 1990; Byrne, 1990). The
Georgia project was credited with initiating subsequent interest in ISP and
served as a model for many others ( Lurigio, 1997). Implemented from

1982 through 1985, the project’s primary goal was to divert nonviolent but

14

serious offenders from prison (Ervvln and Bennett, 1987; Bureau of Justice
Assistance, 1988). The program’s supervision protocol was stringent
relative to standard probation as well as other ISP programs. For example,
during the first phase of a three-phase program, five (5) face-to-face
contacts per week were required and weekly arrest checks were
conducted. Offenders were expected to adhere to mandatory curfews,
meet employment requirements and submit to routine drug and alcohol
screens (Bureau of Justice Assistance, 1988). The project’s results were
promising. Erwin (1987) reported that Georgia’s program played a
significant role in reducing the flow of offenders to prison, noting a 10%
reduction in the percentage of felons sentenced to incarceration. Georgia’s
program appeared to not only be successful in diverting a large number of
offenders from prison, but proponents claimed it produced considerable
cost savings as well. It was estimated that the ISP program saved $6,000
per offender (Erwin, 1990) or the cost of building at least two new prisons
(Erwin, 1986). Further the evaluation reported that participants had lower
recidivism rates and most were able to maintain employment and make
restitution (Erwin and Bennett, 1987).

Second only to Georgia’s program in prominence and replications,
New Jersey’s model is perhaps the best example of a backdoor early

release mechanism to relieve prison crowding (Lurigio, 1990). The

15

program, implemented from 1983 through 1996, targeted offenders
already incarcerated in prison as opposed to diverting offenders from
prison (Bureau of Justice Assistance, 1988). Moreover, the target
population consisted of more offenders convicted of drug sale than did
Georgia’s; 47% verses 24% respectively (Pearson and Harper, 1990) and
placed significant emphasis on meeting the treatment needs of this
population. Conducting offender needs assessments and making referrals
to counseling were ongoing processes. Counseling was described as the
cornerstone of the program (Fulton and Gendreau, 1995).

The admission criterion for the New Jersey program was very
stringent and prolonged. Offenders had to apply within 30 to 60 following
admission into prison, and had to pass through a seven-step selection
process (Pearson and Harper, 1990). In applying, offenders were required
to develop their own program plan to include Identification of a community
sponsor and a plan for employment and housing. Applications were
reviewed by a screening board and if accepted, a recommendation for an
amended sentence was sent to the judge. Much like the Georgia model,
the program called for high levels of contact (Pearson and Bibel, 1986).
The program’s outcomes appeared promising as well. The program
evaluation (Pearson and Harper, 1990) found that ISP offenders, when

compared to parolees over a 24 month period, were significantly less likely

16

to be rearrested and convicted. Moreover, the evaluation claimed that ISP
had achieved an employment rate of over 96%, reduced costs by $6,000 to
$7,000 for each offender per year, saved over 62,000 offerder-days of
prison time and achieved a significantly lower recidivism rate than the
comparison group (Bureau of Justice Assistance, 1988).

The positive results of the initial evaluations of ISP in Georgia and
New Jersey were highly publicized and used to justify expansion of both
programs (Byme, 1990). As such, other states moved to develop similar
ISP programs, e.g., Wisconsin, Massachusetts, Ohio. However,
subsequent evaluations of these programs were less encouraging. Neither
Wisconsin nor Massachusetts reported any cost savings nor did Wisconsin
or Ohio report any reduction in recidivism (Petersilia and Turner, 1993b). It
is repeatedly argued by Petersilia and her colleagues (1990; 1992; 1993a;
1993b; 1997) that these inconsistent findings are due to weak and poorly
designed evaluations that do not permit differentiation between program
and participant effects, thus any claims about the effects of ISP from
evaluations that did not employ randomized designs, are suspect.
“Unfortunately, when these and other evaluations were examined more
closely, it became clear that the effects of ISP remain unknown . . . ”
(Petersilia and Turner, 1993b).

Responding to the need to empirically demonstrate the effectiveness

17

of the ISP programs, the Bureau of Justice Assistance (BJA), US.
Department of Justice provided funding to conduct a large scale
randomized study of ISP programs. The Intensive Supervision
Demonstration project which ran from 1986 through 1991, consisted of 14
programs in 9 states and involved approximately 2,000 offenders. Eligible
offenders were randomized to either experimental programs (lSPs) or
traditional sanctions (either prison, routine probation, parole or other forms
of ISP). The target population had to meet only two criteria. The
population had to be adult offenders and not currently convicted of a
violent crime. Once these criteria were met, each site was free to tailor
their programs to meet local needs. Only two sites chose to develop prison
diversion programs while the others chose probation enhancement or
parole enhancement of the more serious offender (Petersilia and Turner,
1993a). Thus, there were great variations among the 11 randomized field
experiments.

In general, the findings suggest that by providing closer supervision,
ISP programs are effective surveillance and intermediate sanction
programs. Researchers reported:

Most of the ISP's were significantly higher than the control programs

in number of face-to-face contacts with supervisors, telephone and

collateral contacts, law enforcement checks, employment monitoring,

and drug and alcohol testing . . . Most of the ISP's had significantly

higher levels of curtailed freedom . . . The rate of technical violation

was high, making the resultant coercion and diminution of freedom

18

experienced by the offenders an added punitive sanction as well as

creating a public safety benefit (Petersilia and Turner, 1993b).
However, the level of surveillance does not appear to have a positive effect
on offenders' subsequent behavior. Specifically, the data showed that there
was no significant difference in the number of rearrests for new crimes or
the severity of new crimes between the ISP and control cohorts. ISP
participants were not subsequently arrested less often than the control
group, nor did they have a longer time to failure. Another study of over
12,000 adult felons placed on probation replicated these findings and
concluded that there is no association between intensive supervision and
rearrest rates (Langan,1994). With minimal monitoring, regular probation
has no worse recidivism than ISP. In fact, rearrest rates among ISP
participants are somewhat higher due to more technical violations. When
monitoring procedures are intensified, the likelihood that a probation
violation will be identified is greater. On this particular measure, ISP
programs are not very successful (Petersilia and Turner, 1993a).

Like other inquires about ISP's, the question of whether ISP
programs are a cost-saving alternative has an ambiguous answer and is
largely dependent on what alternative ISP is being compared to; e.g.,
incarceration, probation etc. (Latessa, 1986). For example, it cannot be

disputed that incarceration is more expensive than probation or even

19

intensive probation. It is estimated that the average cost of imprisonment
per year, per offender is $12,000 as compared to $4,000 for an ISP
offender (Petersilia,1993a). However, if the term of prison commitment ls
significantly less than the time an offender would be expected to spend in
ISP, the cost of incarceration could be less. Correspondingly, if ISP is
utilized as an enhancement to standard probation then the cost of ISP
compared to standard probation would also be greater. Within the
diversion programs participating in the demonstration project, some ISP
participants spend part of the follow-up year incarcerated anyway, due to
technical violations. This offsets any initial cost-savings of these ISP
programs. Moreover, in the ISP enhancement programs high violation and
incarceration rates drove up the estimated costs for ISP participants to
$7,200 per year compared to $4,700 for the control group (Petersilia and
Turner, 1993a). Researchers concluded (Petersilia and Turner, 1993b):

In general, the evaluation findings show that ISP programs did not
alleviate prison crowding and may have increased it in some states; they
cost considerably more than advocates have realized, particularly if
agencies incarcerate offenders for technical violations and rule infractions;
they are no more effective than routine probation and parole in reducing in-
program recidivism (as measured by arrests and convictions); they did

provide a means by which offenders were held more accountable for there

20

crimes and community behavior; and they may have increased public
safety.

Repeated studies continue to reported similar results (Turner et al.,
1992; Langan, 1994; Petersilia and Turner, 1990). Overall, ISP programs
appear to be effective in providing closer supervision, but have little impact
on subsequent criminal behavior. Moreover, technical violations are
greater among ISP probationers and if violators are incarcerated, ISP
programs can be more costly.

After a decade of experimentation with ISP, the research suggests
that ISP programs are not achieving their stated goals of alleviating prison
crowding, reducing costs or reducing recidivism (Tonry, 1990). It would
appear that all intensive supervision does is monitor offenders’ success of
failure in meeting the conditions of ISP. These results would indicate that
the current emphasis of ISP on surveillance cannot be justified based on
the goal of recidivism (Byrne, 1990). But before “throwing the baby out
with the bath water” other aspects of these studies must be mentioned.
There is sufficient evidence to suggest that offenders participating in more
intensive probation programs are more likely to participate in substance
abuse treatment and counseling. For example, data from the National
Demonstration ISP Project show that 45% of ISP offenders participated in

drug and alcohol abuse counseling compared to only 22% of offenders in

21

the control group. Moreover, this treatment participation was associated
with 10 to 20 percent reductions In recidivism (Petersilia and Turner,
1993a). In programs where offenders received drug testing and
participated in treatment, recidivism was reduced 20-30 % (Petersilia and
Turner, 1997). Recent program evaluations in Texas, Wisconsin, Oregon,
and Colorado have reported similar results (Clear and Braga, 1995). A
study by Langan (1994) also found that if probationers were participating
and making progress in treatment they were less likely to have a new
arrest as compared to drug offenders who were not ordered to be tested or
treated. Given the correlations between offenders participating in
substance abuse treatment programs and recidivism, ISP programs can be

an effective means to the reduction of subsequent crime.

Compulsory Treatment:

The operant term here is “if probationers were participating.”
Treatment cannot work unless the offender accesses treatment or in other
words, unless treatment is administered. For many drug abusers, the
criminal justice system may provide the only contact with treatment. Most
offenders have not been treated in the community and community-based
sanctions can offer an opportunity to engage the substance abuser in

treatment (Tims and Leukefeld, 1992). Because ISP programs have been

22

 

proven effective in providing increased levels of monitoring, ISP programs
can provide an effective context for court-ordered or compulsory substance
abuse treatment.

The underlining premise of compulsory treatment assumes that, of
the numerous types of substance abusers, few are motivated for treatment.
Consequently, some sort of lever for structuring treatment for those who
would not seek treatment or stay in treatment under their own volition, must
exist. This lever has come to be known as "rational authority" and
suggests non-punitive, but mandatory or compulsory, "treatment" (lnciardi,
1988)

The use of “rational authority” or legal sanctions to force offenders
into treatment is not a new concept. Prison based treatment was
proposed for the first time in the United States following the passage of the
Harrison Act of 1914. Prior to the 1920's, the Narcotics Unit of the
Treasury Department encouraged Congress to establish a conglomerate of
Federal "Narcotic Farms" where heroin users could be treated while
incarcerated (Brecher, 1972). Subsequently, specialized treatment for
addicts in the United States began in 1935 with the opening of a Public
Health Service in Lexington, Kentucky and another in 1938 in Fort Worth,
Texas. These hospitals treated addict-patients who voluntary requested

commitment, and involuntary patients, who had been prosecuted for

23

criminal offenses, incarcerated, and civilly committed for treatment
(Leukefeld and Tims, 1988b).

Patients spent 6 to 12 months at these hospitals. The hospitals
followed a standard course of withdrawal, physical restoration,
psychological therapy in the form of group and individual counseling, and
vocational counseling. Following the course of treatment, patients would
be returned to their communities to resume their lives. However, the
physical structure of these "hospitals" resembled a prison with walls, bars,
and strict security and most voluntary patients did not remain in treatment
for the entire program. Relapse rates were extremely high and outcome
reviews were not favorable (Morgan, 1981; lnciardi, 1988).

Critics argued that the Lexington and Fort Worth programs were
almost total failures. Follow up studies showed that between 1935 and
1964 there were 87,000 admission at the two facilities. Of these
admissions 63,600 were considered "voluntary" patients and 23,400 were
Federal prisoners. Of the voluntary patients, seventy percent (70%) had
left against medical advise. Of the total admissions, ninety percent (90%)
had relapsed within a few years (lnciardi, 1988). However, a number of
implementation flaws, not the concept, were acknowledged as barriers in
achieving success.

Building on the Lexington model, and lead by the new treatment

24

based philosophy, a series of new programs were designed. In 1961, the
California State legislature established the California Civil Addict Program
within the Department of Corrections for the compulsory treatment or non-
punitive incarceration of narcotic addicts. During this same period several
cases were being brought before the US. Supreme Court challenging the
Constitutional rights and guarantees of prisoners. The "hands of doctrine"
was beginning to erode and the mood of the Court was changing. In 1962
the United States Supreme Court in Robinson v. Califomia, 370 US. 660,
82 S.Ct. 1417, 8 L.Ed.2d 758 (1962), held that punishing a person for being
a drug addict is a violation of the Eighth Amendment of the US.
Constitution and constitutes cruel and unusual punishment. The rationale
set forth by the Court stated that drug addition was an illness, and therefore
a state could not make addiction a status of crime (Gostin, 1991). The
Court's decision gave additional support to the commitment for treatment
vs. punishment approach and in 1963, the California state legislature
amended sections of the California Rehabilitation Act to further emphasize
the treatment philosophy.

The revision of the California statutes provided for civil commitment
to the California Rehabilitation Center for an indeterminate period for up to
seven years, without first being convicted of a crime (lnciardi, 1988). An

individual in need of treatment could be civilly committed by one of three

25

methods. First, after conviction for a misdemeanor or felony - but prior to
sentencing, a separate superior court proceeding could be held to
determine whether the person was an addict or in danger of becoming an
addict. Second, any "concerned" party could report under oath to the
district attorney (DA) the belief that another person was addicted to
narcotics or was in danger of such. lf probable cause was present, the DA
could petition the superior court for commitment not to exceed seven years.
Lastly, any person who believed themselves to be an addict in need of
treatment could request of the DA a petition to the superior court for
commitment not to exceed two and one-half years. Most of these patients
were involuntary commitments who had been convicted of a crime
(McGlothlin, et al., 1977a). Although the stated intent of the legislation was
"non-punitive", the program was placed under the Department of
Corrections as opposed to a treatment agency, e.g., the Department of
Public Health. Some were skeptical of the legislative "intent" of this action:
...there is little doubt that the political climate in 1961 favored strong
measures to suppress narcotic addiction and that the intent of civil
commitment legislation was at least equally as much for CONTROL
as for treatment (emphasis added) (McGlothlin et al., 1977a).
There seems to be little question that the tensions resulting from the
opposing ideologies between corrections and treatment are long standing.
At the Federal level, Congress enacted the Narcotic Rehabilitation

26

Act (NARA) in 1966 which consisted of four Titles. Titles I, II, III of the act
provided for; civil commitment of those choosing treatment or as requested
by a relative without a criminal charge; those assigned after a criminal
conviction, but in addition those persons charged but not convicted of any
offence could also be committed. Drug offenders could seek treatment as
an alternative to criminal prosecution. Title IV provided for financial
assistance to States and localities for treatment programs (Maddux, 1987).

The NARA authorized the Surgeon General to contract any public or
private agency for the provision of drug treatment services. To expedite
program development and implementation, the Lexington and Fort Worth
hospitals were utilized. These facilities were renamed "clinical research
centers" and NARA admissions to the centers began in 1967. All other
admissions (federal prisoners and voluntary) ceased by 1968. However,
patients that entered the two centers were not much different than those
previously admitted with prisoner or voluntary status, nor was the treatment
much different. Even though the NARA program required new and different
procedures, the fundamental programs of the centers remained. Little
change in response to NARA occurred (Maddux, 1988).

Grants under other legislative authority and grants to states and
other communities under Title IV created increasing local services for drug

abuse treatment. With the increase in local services there was a decrease

27

in Title III admissions: civilly committed addicts not charged with criminal
offenses. Drug Abusers could only be admitted to Lexington and Forth
Worth if State or local facilities were not available. Therefore, the clinical
and research missions of the facilities were lost and in the early 1970's

both facilities were closed (Maddux, 1987, 1988).

Effectiveness of Compulsory Treatment:

The California program proved to be effective at modifying
behaviors. The initial evaluation study of the California Civil Addict
Program (CAP) was performed during 1974 to 1976 and examined the
efficacy of mandatory treatment and civil commitment. Data from 1,000
individuals admitted from 1962 to 1964 for a seven (7) year period of
commitment were selected for follow-up. Eight (8) years of pre-admission
data and 11 to 13 years of post-admission data were obtained during
interviews (Anglin, 1988; Leukefeld and Tims, 1988b). To evaluate
subsequent drug use, researchers examined the percentage of time during
each year that narcotics were used on a daily basis. Data showed that
among the treatment group there was an immediate and dramatic decrease
in daily narcotic use which was sustained over a 5 year period. Moreover,
a sharp and sustained reduction in associated property crimes was also
observed among the treatment cohort. Researchers concluded that civil
commitment is effective in suppressing daily drug use and criminal

28

involvement (McGlothlin, et al., 1977a). Two follow-up studies of the NARA
program (Langerauer and Bowden, 1971; Stephens and Cottrell 1972)
found similar results with respect to drug abstinence. Each reported that
patients treated in the NARA program had better results than patients who
volunteered for hospitalized treatment (Maddux, 1988).

Studies also show a relationship between the length of time in
treatment and treatment outcomes among legally coerced clients. Utilizing
multivariant analysis, studies by Simpson and Sells (1982), Barr and Antes
(1981) and Holland (1983) identified time in treatment as the most
consistent predictor of positive client outcomes. Even among clients who
drop out, the link between length of stay in treatment and treatment
success exists (DeLeon, 1984; DeLeon 1988). One of the largest studies
conducted that demonstrated this relationship is the Treatment Outcome
Prospective Study (TOPS). This large-scale study included 12,000 clients
from 10 cities. From these 10 cities, 41 various publicly funded treatment
programs participated including: outpatient methadone, residential, and
outpatient drug-free from 1979 to 1981. The purpose of the study was to
determine the major factors that affected treatment outcomes (Hubbard, et
al., 1988). The study results not only support the basic belief that criminal
justice clients do as well or better than other clients in drug abuse

treatment, but also legally referred clients stay in treatment longer and

29

length of time in treatment is associated with treatment outcomes.

Other studies have been conducted that compare outcomes with
specific treatment time intervals (Wexler, et al., 1990; Field, 1989). Field's
1989 study evaluated one of the most widely known substance abuse
treatment programs for incarcerated offenders, the Cornerstone program
located in the state of Oregon. Among one of the many study objectives,
the evaluation examined the effectiveness of compulsory residential
treatment and the relationship between length of time in treatment and
recidivism. Researchers compared a group of program graduates with an
average program length of stay of 11 months with 3 groups of offenders
who did not graduate: 1) offenders who spent over 6 months in the
program; 2) offenders who spent 2 to 6 months in the program; and, 3)
offender who were in the program less than 2 months. The groups were
followed for 3 years and each group was evaluated based on the percent of
each group without arrest, without conviction, and without incarceration of
jail terms greater than 6 months. Results showed that more than 50% of
the graduates were not convicted, 75% were not incarcerated, and over
one-third (1/3) or 37% were not rearrested. Of the non-graduates who
were in treatment over 6 months, only 28% were not convicted, 37% were
not reincarcerated, and 21% were not rearrested. Of the drop-outs who

spent less than 2 months in treatment, only 8% were not rearrested in the 3

3O

year follow-up period, 11% were not convicted, and only 15% were not
reincarcerated (Field, 1989). Researchers conclude that time in treatment
correlates positively with measured decreases in criminal activity.
However, more does not necessarily always mean better. In a large
scale evaluation study of New York’s well known Stay’n Out residential
treatment program, researchers sought to examine the effectiveness of
compulsory prison based treatment on recidivism (Wexler, et al., 1990). All
offenders who participated in the program between 1977 and 1984 were
included. One of many elements of the study examined the effect of
specific lengths of time offenders remained in treatment to subsequent
arrest. Program participants were divided into 5 subgroups according to
the length of time they were in treatment: less than 3 months; 3 to 5.9
months; 6 to 8.9 months; 9 to 11.9 months; and greater than 12 months.
The results of the study affirmed findings in other studies that showed
increases in the length of time in treatment to be significantly related to
positive treatment outcomes. Of the offenders who remained in treatment
9 to 12 months, 77% had no parole revocation, no arrests or no convictions
of any kind for 3 years ( Wexler, et al., 1992; US. Cong. House, May
1991). However, the most provocative finding of the study showed a
decline in positive outcomes of offenders who remained in treatment for

more than 12 months. Researchers concluded that there is a optimal

31

duration for treatment and that greater exposure to treatment produces
positive effects only up to a point of satiation. This finding suggests the
need for further research to determine the optimal levels of treatment for
different offender populations under various conditions.

There is more than 25 years of supporting evidence that coerced
substance abuse treatment for offenders is as effective, and in some
cases more effective at producing positive treatment outcomes than
voluntary treatment (Wexler, et al., 1990; Fields, 1989; Langenauer and
Bowden, 1971; Stephens and Cottrel, 1972; Anglin, 1988; De Leon: 1988,
Anglin and Hser, 1990a; Lipton, 1995). Moreover, the relationship between
compulsory treatment and positive outcomes is consistent across the four
major treatment modalities: therapeutic communities; outpatient drug-free;
social models; and methadone maintenance (Gerstein, et. al., 1994).
Studies support the conclusion that how a substance abuser is exposed to
treatment seems irrelevant. What Is important is that substance abusers
are brought into an environment where treatment can occur and that
treatment is long enough to bring about change and recovery. Unfortunately
the recovery process is often misunderstood by criminal justice officials
which can lead to an unrealistic expectation for immediate change in the

drug-involved offender’s drug seeking behavior.

32

The Addiction Process and Recovery:

In order to understand the recovery process it is necessary to
understand the nature of addiction. People who fail to recover from alcohol
and drug addiction do so often because they do not understand their
addiction. It is argued that erroneous information about the nature of
addiction is responsible for much improper or incomplete treatment that
often leads to alcohol and drug relapse (Gorski and Miller, 1986; Martin,
1982). Although theories abound relative to the nature of addiction, most
professionals in the field of addiction science agree that alcohol/drug
addiction is a chronic relapsing incurable but treatable physical disease
characterized by compulsive drug seeking behavior, and should be treated
as such (Ohlms, 1993; Leshner, 1997).

Scientific advances over the past 20 years in the biological and
neurosciences have furthered the understanding of the physiological nature
of addiction. Early research in liver metabolism has shown that people
who develop a high tolerance for alcohol and develop alcoholism
metabolize alcohol quite differently compare to people who drink and never
develop the disease. The hepatic metabolic anomaly in combination with
biochemical processes cause irreversible alterations in the brain’s
chemistry leading to addiction to alcohol and other sedative drugs, referred

to as “sedativism”. Specifically, science has shown that in people who

33

develop alcoholism/sedativism, a small portion of the bi-product produced
from metabolized ethanol (acetaldehyde) combines with the brain’s
neurotransmatter dopamine and biologically an opiate-like addictive
chemical substance is produced in the brain know as tetrahydroisoquinline
(TIQ). Research suggests that this process only occurs in the brains of
individuals who develop alcoholism/sedativism and not in the brains of
people who, even though may drink for years, never develop the disease.
Once produced, TIQ remains a part of the brains chemical composition and
is not eliminated even though the alcoholic may stop drinking for years. It
is theorized that this process may explain the progressiveness and
chronicity of the disease: once addicted, always addicted and long periods
of abstinence will not reverse the process (Cohen, 1978; Hamilton et al.,
1978; Myers, 1978; Ohlms, 1993). Moreover, these metabolic differences
are more often seen in people with a family history of alcoholism and are
present even before there are any indications of problem drinking (Schuckit
and Rayses, 1979). These studies strongly support studies which suggest
a genetic and hereditary basis for addiction (Cloninger et al., 1981; Cotton,
1979; Goodwin, 1980; Schuckit et al., 1985).

The biological and neurological basis for addiction is further
supported by studies which led to the suggestion that people at risk for

addiction to certain drugs may suffer from deficiencies in the brain’s natural

34

chemicals and neurotransmitters that determine mood and serve to relieve
pain and stress: e.g., endorphins, norepinephrine, seratonin and dopamine
(Gold et al.,1985; Khantzian, 1985; Levinthal, 1988). For example, there is
empirical evidence documenting that endorphine levels are abnormal in
opiate addicts (Goldstein, 1978; Ho et al., 1980). According to Levinthal
(1988), a deficiency in an endorphine system that ordinarily would support
feelings of pleasure and reinforcement could lead to feelings of inadequacy
and sadness. It is argued that people at risk for narcotic addiction may
suffer from an impairment of the body’s ability to produce endorhpines and
further, that this deficiency may be genetically determined. Goldstein
(1978) hypothesized that a person could inherit an endorphine deficiency
and if they began to use narcotics they would discover a “normalizing” or
euphorogenic effect in excess of that experience by people without the
abnormality. This effect would predispose a person to opiate addiction and
make it harder for them to remain abstinent. Not only does the endorphine
deficiency, which is observed in the heroin addict, provide a plausible
medical reason for relapse, Schuckit (1980) argues that additional genetic
factors play a role in the physiological drive to return to drugs, as mediated
by a protracted abstinence syndrome or through inherited psychological
vulnerabilities.

Once addicted, Leshner (1997) argues that more recent scientific

35

evidence shows that the addictive brain becomes even more distinctly
different than the nonaddictive brain in that all drugs of abuse have
common effects on a single pathway within the brain referred to as the
mesolimbic reward system. Activation of this system appears to be a
common element in addiction or what keeps drug users taking drugs and is
not unique to any one drug. All addictive substances appear to affect this
circuit in the same way. Prolong drug use then causes changes in the
brain’s metabolic activity, receptor availability and responsiveness to
environmental cues that can persist long after the person stops taking
drugs.

The implication in understanding the fundamental differences and
changes in brain function as a consequence of addiction for treatment and
recovery is self evident. Even though initial alcohol and/or drug use comes
about as a voluntary behavior, addiction is not voluntary. The addict’s
metabolism and brain chemistry is different from the non-addict’s and the
addict must be dealt with as if she or he has a chronic relapsing illness with
long lasting physical and social consequences requiring long periods of
time for recovery. Recovery from the long-term effects of addictive drugs
on the central nervous system (CNS) alone requires a significant amount of
time. For example, CNS damage or brain dysfunction has been

documented in as much 75 to 95 percent of recovering alcoholics (Abbott

36

and Gregson, 1981). Recovery from CNS damage usually requires 6 to 24
months (Gorski and Miller, 1986). During this period of time, post-acute
physical withdrawal symptoms (PAW) may continue lasting months and
include, cognitive and memory disturbances, excessive anxiety, lrritability,
insomnia, depressive symptoms and more. These symptoms of long-term
toxicity and withdrawal associated with CNS damage can take up to 18
months to stabilize and are often related to relapse (McGrady and Smith,
1986; Porjesz and Begleiter, 1983). The recovery process can create a
great deal of stress and many chemically dependent people have difficulty
Ieaming to manage stress without the use of alcohol or drugs. It is
suggested that the stress associated with recovery aggravates the brain
dysfunction and exacerbates the long-term withdrawal symptoms often
leading the addict back to drugs to relieve the symptoms (Gorski and
Miller, 1986). However, with abstinence, treatment and time these
symptoms will disappear and recovery is made possible.

Time is a critical element in reversing the physiological, sociological
and emotional deterioration and dysfunction of the addict. In addition to the
time necessary to complete long-term withdrawal and repair neurological
damage, some researchers suggest that the most serious problems caused
by addiction require 2 to 3 years to resolve (DeSoto et al., 1985).

However, recovery should not be conceptualized as a developmental

37

process of simply repairing or reversing damage created by the addictive
disease. In addition, successful recovery demands the development of
entirely new coping skills, the development of a new personal identity and
approaches to living. Recovery goes beyond “re-covering” what was once
lost to the acquisition of a new lifestyle. The more longstanding lifestyle

changes require 8 to 10 years for full resolution (DeSoto et al.,1985).

The Need for Program Evaluations and Change:

Despite research documenting the need for substance abuse
treatment and the effectiveness of compulsory treatment most ISP
programs are unable to ensure compliance with court-ordered substance
abuse treatment. As a result, these programs are not as effective as they
could be. It is argued that many ISP program models are plausible and
could work except they lack the necessary treatment resources (Petersilia,
1997). Because ISP practices generally center around surveillance and
enforcement activities and are not treatment focused, programs are often
implemented without creating the organizational capacity or resources to
provide the necessary substance abuse treatment. This apparently reflects
the values of policy makers. It appears that most program policies tend to
support punishment of the offender as opposed to treating the drug-

involved offender as having a chronic relapsing disease characterized by

38

“drug-seeking behavior.” Therefore a significant proportion of offenders
who are in need of treatment or court ordered to treatment never receive
treatment services. Langan ( 1994) found in a study of 12,370 felony
probationers, 24% of those ordered to alcohol treatment and 32% of
offenders ordered to drug treatment never received any treatment by the
end of their probation term.

In order to provide for court-order treatment, the focus of ISP
programs needs to change. Fulton and Gendreau (1995) argue that ISP
programs need to change the way in which they are intensive by shifting
the emphasis away from incapacitation and punishment toward a more
integrated approach of intervention and substance abuse treatment. Given
what is known regarding the nature of addiction, punishment without
treatment is futile. Research has consistently shown that successful
programs combine both treatment and surveillance and target an
appropriate offender subgroup (Petersilia and Turner, 1993a). However, it
is unknown what levels of both treatment and surveillance are required to
maximize outcomes and reduce drug-abuse behavior and recidivism. Is
there an optimal length of treatment time as suggested in the Stay’n Out
study? Will different combinations of treatment and surveillance produce
different results in different population? By modifying and retesting various

program models, the chances for improving the effectiveness of ISP

39

programs will increase. Moreover, with continual budget constraints, it is
imperative that the many unanswered questions concerning supervision
and treatment are explored in efforts to realize and maximize the potential
of probation. More support must be gathered that convinces both criminal
justice professionals and the public that community sanctions are punitive,
will hold offenders accountable for their behavior, produce positive
outcomes; and whether ISP, a concept sound in theory, can be structured
differently to produce more effective results.

This study responds to the question of whether a different and unique
structure of ISP can lead to different outcomes. As such, the primary aim
of this evaluation research is to determine whether meaningful differences
in outcome measures (treatment access, length of time in treatment,
subsequent drug use and recidivism/re-arrest) exist between drug-involved
felony offenders who were sentenced to residential probation and ordered
to outpatient substance abuse treatment, and similarly situated offenders
sentenced to less intensive probation and also ordered to outpatient
substance abuse treatment. The residential probation program selected
for this outcome evaluation is of particular interest in that it mitigates key
issues identified in the literature as barriers to successful ISP programs.
By including on-site substance abuse treatment, the residential probation

program provides the organizational capacity to treat all offenders in need

40

 

 

 

of substance abuse treatment, and offers a greater emphasis in the
integration of treatment and punishment. The following section describes
the residential program in detail and the environmental context of the

research setting.

41

 

III. THE RESEARCH SETTING

Overview:

Intensive supervision probation programs for drug-involved felony
offenders focus on two activities designed to reduce recidivism and drug
use. These activities are: 1) substance abuse treatment and 2)
surveillance or monitoring. In general, there is evidence supporting the
contention that more intensive, treatment-oriented programs could produce
different results in some populations (Petersilia and Turner, 1993a). The
residential probation program selected for this study is only one of many
sentencing alternatives for drug-involved felony offender. The County
represented in the research setting utilizes a variety of community based
sentencing alternatives each characterized by differing levels of monitoring
and substance abuse treatment (e.g., standard probation with outpatient
substance abuse treatment, residential probation with outpatient substance
abuse treatment, electronic monitoring with outpatient substance abuse
treatment). It is therefore possible to track similarly situated offenders
sentenced to different ISP models within the same jurisdiction to determine
if there are meaningful differences in outcome measures between
offenders sentenced to the various models.

Before discussing the study design, it is important to first describe

42

aspects of the research setting. These aspects include the process in
which probationers may be sentenced to residential probation, the
residential probation program intervention and the substance abuse

treatment program.

Environmental Context:

The residential probation program selected for evaluation is located
in a Mid-westem county with a population of 430,459. Of the county’s total
population, 47.9% are males and 52.1% are females. The median age is
32.0 years with 43.2% of the population between the ages of 18 to 44
years. The ethnic distribution is 78.2% white, 19.6% black and 2.2% other
(Table 1).

In 1996, approximately 1,403 felony offenders were sentenced in the
County’s Circuit Court. Of these offenders, approximately 730 offenders2
were sentenced to one of various models of probation; standard probation,
electronic monitoring, boot camp, residential probation; or a combination of
any of the models. The probation model an offender is sentenced to is
affected by many factors during the pre-sentencing process. Prior to

sentencing, a probation agent conducts a pre-sentence investigation (PSI)

 

2

This information was provided by the Office of Community Corrections and obtained
from the Basic lnforrnation Reporting database maintained by the county in the research setting.
The database does do not include mandatory and escape sentences. To protect the anonymity
of the research setting, this source is not referenced.

43

 

 

 

 

Table 1 Population and Demographics of the Research County3
Gender N % A e N %
Males 206,003 47.9 < 5yrs 33,436 7.8
Females 224,456 52.1 5 - 17 87,227 20.2
Total 430,459 100 18 - 20 19,906 40.6
21 - 24 24,484 5.7
Ethnicity N % 25 - 44 137,491 31.9
White 336,651 78.2 45 - 54 46,425 10.8
Black 84,257 19.6 55 - 59 19,676 4.6
Other(Asian, Hispanic, 9,551 2.2 60 - 64 17,985 4.1
American Indian, Eskimo, 65 - 74 26,231 6.2
Pacific Islander, all other) 75 - 84 13,442 3.1
> 84 4,156 1.0
Total 430,459 100 Total 430,459 100

 

and makes recommendations to the court regarding sentencing. One of
the county’s 32 probation agents responsible for conducting PSI’s is
assigned to the case. This assignment is based solely on the current case
load of all 32 agents. In conducting the PSI, the probation agent uses
objective criteria as well as his/her individual assessment skills to
determine if an offender is likely to be eligible for residential probation.
Once this determination is made, the decision to recommend an offender
for residential probation is discretionary. Program administrators recognize
that among the cadre of probation agents some may be more likely than
others to recommend residential probation, but others do not utilize this

option at all. One official acknowledged, “sometimes it’s a fine line and a

 

3

This information was provided by the county’s Metropolitan Planning Commission. To
protect the anonymity of the county and research setting, the source is not referenced.

44

 

toss-up where we recommend they go.” The discretion of the probation
agent is, therefore, an important element in determining whether a
recommendation is made to the court for residential probation for offenders
who may be eligible. Thus, not all probationers who meet the criteria for
residential probation are recommended for the program.

The process in which cases are assigned to circuit court judges in
this jurisdiction is similar and equally as important in determining
placement in residential probation. In the circuit court represented in the
research setting, cases are assigned to any one of seven4 sentencing
judges based exclusively on “blind draw”5. This suggests, in theory, that all
cases have an equal probability of being assigned to the docket of any one
of the presiding judges and moreover, that the distribution of cases for each
judge is similar. However, this was not empirically documented. Judicial
discretion and sentencing practices can also affect whether an offender
who may be eligible for residential probation actually receives this
probation option as a sentence. Based on preference and support of the

residential probation program, some judges may override PSI

 

In 1996, the county’s circuit consisted of 7 judges. In the study cohort, it was determined
that cases were disposed of by 10 different judges. During 1996, 3 judges were replaced
accounting for the variance.

5

This information was provided by a presiding judge in the county’s circuit court during a
personal interview. To protect the anonymity of the jurisdiction and programs, the interview and
the name of the judge are not referenced.

45

recommendations and refer to residential probation even when another
form of probation is recommended. For example, in the study cohort, 69
convicted felons were sentenced to residential probation. However, in only
27 cases were recommendations for residential placement made by
probation agents in the pre-sentence investigations. And of the ten
sentencing judges, three judges accounted for 58% of the sentences to
residential probation. It is also possible that some judges are more or less
likely to sentenced felons to a particular type of probation program based
on offender characteristics such as age, race, prior felony convictions etc.
In order to control for these potential variations in sentencing patterns
several offender characteristics were included in both the Cox regression
and standard logistic regress models as independent variables when
testing the hypotheses.

The likelihood that offenders who are both eligible for residential
probation at the beginning of the criminal justice process and who actually
receive this sentencing option is affected by the random selection of both
the PSI investigator and the sentencing judge. Thus, it could be argued
that the probability of any offender who meets the admission criteria for
residential probation actually being sentenced to residential probation is
affected by the random assignment of both the PSI investigator and

sentencing judge, as well as judicial discretion.

46

 

The Intervention Program:

The residential probation program serves as a sentencing alternative
for nonviolent prison and jail bound adult male offenders and provides an
intense level of supervision for a period of up to six months. Within the
residential probation program, several different programs exist. Each is
designed to target specific offender populations. For example, the
residential program provides a “Long-Tenn Alcohol” program to target
offenders convicted of a third “Operating Under the Influence of Liquor”
(OUIL) offence. In addition, a 30 day “Probation Rule Violation” program is
designed for technical rule violators, a 30 day after care program for
offenders released from boot camp and, the “Standard Program for Circuit
Court” (SPCC) provides longer term services for felony offenders. It is the
SPCC residential probation program that is the focus of this evaluation
research.

To be eligible for admission to SPCC, the offender must first be male
and at least 17 years old. He must be charged or convicted of a felony
crime that carries a minimum sentencing guidelines of “0 to 9 months " or
convicted of a probation violation while on probation for on a felony crime
with a sentencing guideline of “0 to 6 months". In addition the offender

must not:

47

. present a continuing pattern of aggravated assaultive

behavior,

. be recently or currently addicted to intravenous drugs,

. require medical alcohol detoxification,

. present a pattern of debilitating mental illness,

. be currently suicidal or demonstrate suicidal behavior,
and;

. have any serious medical problems that would preclude

active participation in the program, or suffer from a life-
threatening communicable disease.
The offender also must have a functional I.Q (educable) and able to speak
and understand the English language.

The SPCC includes three graduated levels of supervision, the first of
which provides the most intense level of monitoring and is the most
restrictive for the offender. As the offender demonstrates progress and the
ability to follow the program’s rules, he is rewarded by graduating to the
next level until he reaches the third level and is ready for discharge. He
may also receive “credit days” towards early discharge for compliance,
work or for attending classes. While in residence, SPCC participants must
adhere to strict supervision and behavior standards and participate in

educational programs and life skills groups. All activities and the offender’s

48

whereabouts are closely monitored. Any rule violation is subject to
disciplinary consequences ranging from loss of “credit days” to probation
revocation depending on the severity or frequency of the violations.
Routine and random urinalysis and breathalyser tests are conducted, as
well as personal and facilities searches. Canine drug teams are
periodically invited to check the building for drugs. Offenders may be
permitted work release at approved work sites. The average length of stay
in SPCC is four months.

The SPCC population includes many substance-abusing offenders.
For this population, the residential probation program also provides on-site
substance abuse treatment. Although the treatment is provided in-
residence, the treatment is not based on a residential treatment or
therapeutic community model, but rather on an out-patient treatment model.
Thus, the residential probation program integrates the philosophies and
characteristics inherent in intensive probation“) the concepts of specific

deterrence theory7 and the knowledge generated by research in

 

Thompson (1987) describes intensive probation as having 5 elements in addition to
intensive monitoring or frequent contacts. First, supervision is focused and aimed at regulating
specific behaviors e.g., drug use, travel, employment, etc. Second, the offender is usually
subjected to random drug testing. Third, supervision is graduated whereby offenders are
monitored more closely at the beginning and less frequently over time. Fourth, supervision is
strictly enforced and noncompliance is swiftly punished. Lastly, supervision is monitored by
agents who are specially trained and part of an autonomous unit of probation. All of these
elements are visible components of the residential probation SPCC program.

7

Specific deterrence is concerned with changes in the behavior of the convicted offender
(after-the-fact) and punishment or the “threat” of punishment is intended to discourage the

49

compulsory treatment“.

The substance abuse treatment is contracted through a local
substance abuse treatment provider. The outpatient substance abuse
program consists of a two-phase process lasting an average of fourteen
weeks. During the first phase of treatment, offenders are required to
participate in two sessions per week for six weeks. In the second phase,
the offender participates in two sessions per week for eight to nine weeks.
While in treatment, each probationer is assigned to a probation case
manager who assists in identifying the individual needs of the offender and
facilitates appropriate referrals to outside human service agencies when
appropriate.

Once the probationer has completed the in-residence probation and
substance abuse treatment programs, offenders typically return to standard
probation, where probation officers monitor the offender's behavior, check
re-arrest status and conduct urinalyses. At this point, the level of

monitoring is determined at the discretion of the probation agent. If

 

convicted from committing crimes in the future. The threat of additional legal sanctions,
revocation of probation/incarceration, is used to discourage specific behaviors and increase
compliance. In contrast, general deterrence holds that the population at large is dissuaded from
criminal when it sees that punishment follows criminal behavior. The punished criminals are
intended to serve as an example (Clear and Cole, 1994).

8

As previously mentioned, there is more than 25 years of supporting evidence that
coerced substance abuse treatment for offenders is effective (Langenauer and Bowden, 1971;
Stephens and Cottrel, 1972; Anglin, 1988; Anglin and Hser, 19908; Lipton, 1995) and moreover,
that treatment is effective regardless of the modality (Gerstein, et. Al., 1994). SPCC requires that
all offenders with substance abuse history’s participate in out-patient treatment.

50

indicated by the treatment staff, the probationer is also referred for
continued outpatient substance abuse treatment. It is recommended that
all probationers continue attendance at Alcoholics Anonymous and/or
Narcotics Anonymous.

The residential probation program selected for this evaluation study
is only one of many sentencing options available to circuit court judges in
the research county. The program provides a unique treatment oriented
model that emphasizes the integration of treatment and punishment. With
a greater emphasis on treatment the program is designed to address the
treatment needs of all offenders who are sentenced to the program and in
need of treatment. All offenders who meet the program’s admission criteria
are eligible for residential probation placement. However, several factors
influence whether offenders are actually sentenced to residential probation
or to another probation model: the random process of assigning PSI
agents; the recommendations of the PSI agents; the random process of
assigning cases to judges’ dockets; and the discretion of each circuit court
judge. Therefore it is possible to track similarly situated offenders
sentenced to different ISP models within the same jurisdiction to evaluate
the effects of residential probation and compare differences in outcome
measures to other ISP models. The following section details the research

design and methods for evaluating the residential probation program.

51

IV. METHODS

Introduction:

Based on previous research discussed in the literature, the primary
research question this program evaluation intended to answer is: “Are there
meaningful differences in treatment access, length of time in treatment,
subsequent illegal drug use and recidivism between offenders sentenced to
residential probation and out-patient substance abuse treatment (SSPC)9,
and similarly situated offenders sentenced to less intense levels of
probation and outpatient substance abuse treatment?” Given the
documented link between substance abuse treatment and the reduction in
subsequent criminal activity, the study was designed to answer questions
directed towards evaluating the residential probation program’s
effectiveness in ensuring compliance with court ordered substance
treatment and to measure subsequent drug use among drug involved
felony offenders. The analysis to examine recidivism is affected by the
reduced sample size for which data was available. This presents

limitations in drawing meaningful conclusions from the measures of

 

9

As previously discussed, the residential probation program has within it several individual
programs designed to target specific populations. This study focuses only on offenders
participating in the SSPC and SSPC/PES programs. Henceforth, the term “residential
probationers” will refer only to participants in the SSPC and SSPC/PES programs.

52

 

 

recidivism. However, measures of compliance with court ordered
substance abuse treatment and drug relapse are good proxy measures for
recidivism. Therefore, treatment access, length of time in substance abuse
treatment and length of time to subsequent drug use are used as the
dependent variables in the study’s main evaluation models. The length of
time to rearrest (failure) for technical violations and new crimes are also
examined. The following section details the selected evaluation questions,

rationale and hypotheses.

Research Questions and Hypotheses:

Evaluation
Question 1: Are drug-involved offenders sentenced to residential

probation more likely to receive substance abuse
treatment than drug-involved offenders sentenced to
non-residential probation models?

Research consistently shows compulsory treatment to be an
effective means in achieving positive treatment outcomes (Maddux, 1988;
Hubbard, et al., 1988; McGlothlin, et al., 1977a; 1977b). Ensuring
compliance with court ordered treatment is therefore, the first step toward
achieving this goal. All probationers in the study sample were required to
participate in substance abuse treatment as a condition of probation. For
the residential probation population, the required treatment is conducted

53

on-site while the probationer is in residence. The program also provides
24 hour intense monitoring and conducts frequent random drug screens.
Offenders are subject to severe legal sanctions if they are found to be
using drugs or do not comply with all rules, inclusive of treatment
participation. It is assumed that offenders who are faced with legal
sanctions (legal threat) if they fail or abscond are more likely to comply with
court ordered treatment mandates, and that the legal threat for not
participating in treatment is greater for probationers within the residential
program compared to probationers in non-residential probation programs.
Given these assumptions, the following is hypothesized.

Hypothesis 1: Probationers sentenced to residential probation

are more likely to participate in substance abuse treatment.

 

Evaluation
Question 2: Are there differences in the length of time drug-involved

offenders stay in treatment and remain drug free
following sentencing, between probationers sentenced
to residential probation and probations sentenced to
non-residential probation models?
As previously discussed, studies find that clients who complete
treatment have better outcomes compared to those who drop out of

54

treatment. For many studies, completion rates are measured as a
dichotomous variable: either the client completed treatment or not. This
measure yields little information regarding the length of time a client
remains in treatment. Several studies well document that the “length of
time” a client remains in treatment is the most constant and best predictor
of treatment outcomes. This is true even among clients who dr0p out of
treatment (Simpson and Friend, 1988; DeLeon, 1988, 1994; Hubbard et
al.,1988; Barr and Antes, 1981). Given the importance of this link, this
study also examines the time in treatment for all offenders who entered
treatment.

Studies also show a relationship between treatment retention and
enhanced case management services (Collins and Allison, 1983; Cook,
1992). Clients receiving case management services have been found to
remain in treatment, both residential and outpatient, up to 7 weeks longer
than clients not receiving case management services (Hubbard, et al.
1988). These augmented services are also provided by the residential
probation program. Each probationer is assigned an in-house case
manager to bridge the criminal justice and treatment systems and facilitate
linkages to ancillary services, (e.g., medical care, social support, shelter,
food, job assistance, etc.). The evidence in the literature and the

residential program’s activities supports the following hypotheses.

55

 

Hypothesis 2a: Drug-involved offenders sentenced to residential
probation will remain in substance abuse treatment longer.
Hypothesis 2b: Drug-involved offenders sentenced to residential

probation will remain drug-free longer.

 

Evaluation
Question 3: Are there differences in the likelihood of rearrest over

time between drug-involved offenders sentenced to
residential probation and offenders sentenced to non-
residential probation models?

This question is answered separating rearrests for technical
violations and rearrests for new felony offences and examining failure over
time. Recent evidence shows the rearrest rates among probationers in
more intensive probation programs are somewhat higher due to more
technical violations. When monitoring procedures are intensified, the
likelihood that a probation violation will be identified is greater (Petersilia
and Turner, 1993b). Moreover, the level of surveillance does not appear to
have a positive effect on rearrests for new crimes cases, the severity of the
new crime or the length of time to rearrest (Langan, 1994; Turner et al.,
1990; Petersilia and Turner, 1990). With minimal monitoring, regular

probation has no worse recidivism than more intense forms of probation,

56

except when substance abuse treatment is involved. When substance
abuse treatment is involved, a reduction in recidivism of 10 to 20 percent
has been found. In programs where offenders received drug testing and
participated in treatment, recidivism was reduced 20 to 30 percent
(Petersilia, 1997). Given these research findings, and the services and
level of monitoring observed at the residential probation program, it is
hypothesized that the risk of rearrest for probation violations will be greater
among residential probationers. It is also hypothesized that the risk of
rearrests for new felony crimes will be lower.

Hypothesis 3a: The risk of failure over time for technical

violations is greater among offenders sentenced to residential

probation.

Hypothesis 3b: The risk of failure over time for new felony crimes

over time is less among offenders sentenced to residential probation.

 

Evaluation
Questionfi; Of the probationers rearrested for new crimes, are there

differences in the severity of the new offence between
residential probationers and non-residential

probationers?

57

It is known that rearrest rates among probationer in more intensive
probation programs are more likely to be higher due to more technical
violations. Recidivism rates and patterns for new felony crimes, however,
are of greater concern. Here only arrests for new felony crimes are
considered and differences in the severity of the new crime are examined.
As previously mentioned, the level of surveillance does not appear to have
a positive effect on rearrests for new crimes or the severity of the new
crime, except when offenders are involved in substance abuse treatment.
Because it is expected that offenders sentenced to residential probation
are more likely to receive substance abuse treatment, and treatment is
related to better outcomes, the following is hypothesized.

Hypothesis 4: Of the probationers rearrested for new crimes, the

severity of the new offence will be less for offenders sentenced to

residential probation.

Research Design and Study Population:

In an ideal evaluation circumstance, offenders would be randomly
assigned to the different forms of intensive probation and causal inferences
about the impact of each program would be drawn from a comparison of
post-sentence behavior. Because this procedure would ensure that

individuals with reasonably equivalent characteristics were sentenced to

58

both SPCC or non-residential probation, it would increase the confidence
that any observed differences in post-sentence behavior could be attributed
to the type of program experienced by the offender. By utilizing a quasi-
experlmental design the strongest causal inferences cannot be drawn.
However, it must be kept in mind that evaluation research is an effort to
obtain “reasonable evidence” that specific activities or services have some
meaningful impact in a specific environment or setting. Randomized field
experiments represent only one of a variety of possible effective impact
evaluation strategies (Rossi and Freeman, 1982; Weiss, 1972; Mohr,
1988). If evaluation research is to be used as a valuable tool for policy
makers, then in environments where randomized designs are not possible,
quasi-experimental designs are the only option.

Although random assignment may be the best experimental method
for testing the differences, due to various environmental constraints it was
not possible for this program evaluation. Instead, this study is designed as
a “prospective observational or cohort study” and uses existing data for
conducting secondary data analysis. In observational or cohort studies
there is no control over the assignment of the exposed experience or the
interventions. Only the individual exposures experienced by each
participant in the study can be observed. Generally, a sample of the

population is selected and information is obtained to determine which

59

individuals have the particular characteristics that are suspected of being
related to the event being investigated. In this study, all offenders meeting
the admission criteria for residential probation were selected. Individuals
are then followed over time and outcomes are assessed through the use of
existing data. In observational studies, the necessary data are often
obtained indirectly by reviewing existing records and consist of information
about the exposure status of the individual and whether the event of
interest occurred after the exposure (Lilienfeld and Stolley, 1994). This

study relies primarily on data in probation files.

The Sample:

The primary aim of this evaluation research was to determine
whether meaningful differences in outcome measures exist between drug
involved felony offenders who were sentenced to residential probation and,
similarly situated offenders sentenced to other forms of probation.
Therefore, it was deemed appropriate to include in the sample only
offenders meeting the eligibility criteria for admission to the residential
probation program. As such, the starting point in identifying the study
sample was to identify all felony offenders who met the eligibility criteria for
admission, but who were not necessarily sentenced to, the residential

probation program during a defined time period. A database of all felony

60

offenders sentenced in the research county is maintained by the
Department of Community Corrections. From this master database, a
subset of data was obtained consisting of only offenders sentenced during
1996 for two reasons. First, because the residential probation program
being evaluated only accepted male offenders, the ability to identify male
offenders in the data set was necessary. Data prior to 1996 did not include
gender codes. Therefore, it was not possible to discriminate male offenders
from female offenders in records prior to 1996. Second, urinalysis data
were needed to measure offenders’ subsequent drug use. This information
is collected and maintained by the county’s adult probation department.
The department’s policy regarding drug testing changed in early 1997 and
some probationer’s were tested less frequently than others during calender
year 1997. Because validity of the outcome measure “subsequent drug
use” is affected by the frequency of drug testing for all probationers, a time
period following 1995 that would yield the greatest and most consistent
information regarding drug testing was identified. Thus, 1996 represented
the most favorable period of time for identifying the study cohorts.

From the 1996 master database, all offenders meeting the eligibility

criteria for the standard residential probation program and who were

61

sentenced to probation1o were extracted. These criteria included: all males
with histories of alcohol and/or drug abuse; convicted of felonies with
sentencing guidelines of “0 to 9" or greater; and who were diverted from
prison and sentenced to probation. In this initial data cut, 205 cases were
identified. All cases had 1995 or 1996 docket numbers. Next, because the
residential probation program also includes offenders sentenced for
technical violations with sentencing guidelines of “0 to 6" or greater on the
original offence, a second set of data was extracted from the same 1996
master data set. This extraction included all 1995 and 1996 dockets of
male felony offenders with histories of alcohol and/or drug abuse and who
were convicted of technical violations and continued on probation. From
this second data extraction an additional 83 cases were identified“. A total
of 288 felony offenders sentenced to probation in 1996, and meeting the
eligibility criteria for residential probation, were identified.

Once the sample was identified, each offender’s probation file was
reviewed to determine the actual probation sentence. Upon review, it was

discovered that offenders meeting the criteria for placement into residential

 

10

In the county’s master data set the sentence type is coded as either, Prison, Probation or
Split. It was not possible to determine which form of probation an offender was sentenced to until
each record was reviewed.

11

Sentencing guideline are not applicable to technical violations. It was not possible to
determine whether these cases met the sentencing guideline admission criteria for the residential
program until each file was reviewed.

62

probation were actually sentenced to one of five different probation models:
Standard Probation; Residential Probation; Electronic Monitoring (PES);
Residential Probation with Electronic Monitoring; or Boot Camp and
Electronic Monitoring. These separate distinctions represent the study’s
five separate cohorts for comparison. During the review process 79 cases
were excluded for the following reasons: 48 files could not be located
during the review period; in 11 cases it was determined that the offender
did not meet the admission criteria to residential probation (e.g., history of
suicidal tendencies, inappropriate sentencing guidelines, etc.), 8
probationers were sentenced to residential substance abuse treatment
instead of out-patient treatment; 4 cases were transferred out to other
county’s for supervision; 3 cases had requirements of only mail-in-
monitoring; and in 5 cases offenders were sent to prison for violations
committed in jail while serving a “split sentence”. In these last 5 cases
offenders never spent any time on probation or were never “exposed to the
risk of failure.” The final sample (N= 209) is distributed among the five
study cohorts accordingly: Standard Probation N = 70; Residential
Probation N = 38; Electronic Monitoring (PES) N = 54; Residential
Probation with Electronic Monitoring N = 31; and Boot Camp with

Electronic Monitoring N = 16.

63

Variables and Measures:

All of the data used in the evaluation were previously collected by
either the county’s department of community corrections, adult probation,
the residential probation program or the substance abuse treatment
provider. As previously discussed, the Office of Community Corrections
maintains an electronic database that includes the data elements required
for state reporting. This information is obtained from the offenders pre-
sentence investigation and court records and collected by the department
of probation. Data from the Office of Community Corrections were
extracted electronically and merged into the evaluation data set. No
attempt was made to alter the operational definitions or coding scheme of
these variable, except in cases where data categories needed to be
collapsed due to the size of the evaluation data set. Data elements that
were necessary but not Included in the electronic file were abstracted from
paper files at either the residential probation program, the adult probation
department, or the substance abuse treatment provider. This process had
advantages in that, in some instances, coding schemes were not
predefined and could be developed to conform to the evaluation design.

The primary source of data came from the offenders’ probation files
which are maintained at the county’s probation department. Although most

of the active case files provided many of the data elements needed for the

64

evaluation, several of the inactive case files were void of information.
Wherever possible, attempts to rectify these omissions were made through
contacts with the treatment provider, the residential probation program or
by review of public records. The following describes the data and variables

used in the evaluation research.

Independent Variables:

Several risk factors are identified in the literature as predictors of
recidivism. Among these include, age, race, employment status, drug use,
prior felony convictions, offence type, employment and treatment retention
(Langan, 1994; Morgan, 1993; Gennaro, 1987; Prichard, 1979; Hubbard et
al., 1998; Maddux, 1988). These and other variables found to be
significant in the bivariant and univariant analyses of this study were
identified as the study’s key independent variables.

Demographic Variables: Ethnicity and age had previously been
recorded in the master electronic database and distinguished as white,
black, Asian, Arabic, Indian, Hispanic and other. Because only one
Hispanic and one Asian were identified in the evaluation data set with all
others reported as either white or black, the two racial categories, white
and non-white were used. Age was previously recorded in two ways. First,

the offender’s actual age at the time of the offence was recorded. Second,

65

age was also assigned an age category or range. For analysis, it was
decided to use the offender’s actual age instead of the age category
because the actual age would yield a more robust and precise analysis.

Education Level: Education level was also recorded in the master
database in seven separate categories. Due to the size of the evaluation
data set and the selected methodology, education level was collapsed into
two categories for analyses; “0-11 grade” and “GED, 12th grade and
beyond 12th grade”.

Employment end Ineome: Employment and income were previously
recorded as binomials. Offenders were reported to either be employed or
not employed at the time of arrest, with incomes as either, greater or less
than $75 per month.

Prior Felonies: Prior Felonies were obtained from the master
electronic data file and recorded as a continuous variable.

Offence Code: This variable was pre-recorded into six categories:
property, drugs, weapons, csc, assault, and arson. Due to the size of the
study population and the selected methodology, offence codes were
collapsed into three categories; property, drugs and violent offences.

Probation Violation: This variable denotes whether the critical
incident which led to offenders receiving new sentences in 1996, was either

the result of a technical violation while on probation for a previous 1995 or

66

1996 felony conviction, or the result of a new felony conviction.

Drug of Cheice: Drug of choice is the drug preferred or primarily
used by the offender. This information is recorded at the time of the pre-
sentence investigation. Probationers are asked by the pre-sentence
investigator which drug they prefer or used most frequently prior to their
arrest of the incidence offence. The information recorded in the pre-
sentence investigation is usually report as: Alcohol; THC; Crack/Cocaine;
Heroin; Tranquilers; or Poly-Drug Use. Poly-Dug use is recorded if the
offender reports more than one drug of choice. From this report, the drug
of choice was abstracted, coded and recorded in the evaluation data set for
each offender as either; Alcohol, THC, Crack/Cocaine, or Poly-Drug Use.
Due to the low frequency of reported heroin as the drug of choice (N=1) this
case was included in the poly-drug use category. Research indicates that
as high as 90% of heroin users also abuse cocaine and alcohol. Therefore
it is reasonable to consider heroin users as poly-drug users (Abadinsky,
1989). In only three cases tranquilers were reported as the drug of choice.
The use of “tranquilers” by definition assumes the use of multiple drugs
within the same drug classification (sedatives) all with similar
pharmacological response. Thus these cases were also included in the

poly-drug use category.

67

Additional Independent Variables:

Jail Days: Many probationers receive a “split sentence” which is a jail
term in addition to probation. It seemed reasonable to assume that a split
sentence itself Is a unique intervention which could impact client outcomes.
Due to the small size of the sample, instead of treating the split sentence
as a separate type of intervention, the length of time offenders Spent in jail
was controlled for as a separate independent variable in the regression
analyses. Therefore, “jail days” represent the length of time each offender
spent in jail after the critical incident arrest and any time served after
sentencing for a split sentenced. This was calculated from the information
found in the probation files.

Study Group: The model of probation each offender was sentenced
to was determined through a review of the probation files. Five separate
probation models were identified: Residential Probation and Electronic
Monitoring; Residential Probation only; Boot Camp and PES; Electronic
Monitoring only (PES); and Standard Probation only. The models were
then ranked ordered from high to low in terms of the frequency of
monitoring expected for each model. For example, unless they are
working, offenders sentenced to PES are expected to report for face-to-
face contacts with their probation officer once a week while on tether. lf

working, the expectation is that they will have face-to-face contacts no less

68

than twice a month. Offenders sentence to residential probation or boot
camp are usually monitored after discharge at the same level as PES for
approximately 90 days. After 90 days, the level of monitoring is
determined at the discretion of the probation agent. For standard
probation, the level of monitoring ranges from 1 to 4 times per month
depending on the offenders individual risk level classification (maximum,
medium, or minimum risk). Considering these practices, Residential
Probation with PES is considered to be to be the most intensive model of
probation, followed by Residential Probation only, Boot Camp with PES,

PES only, and Standard Probation.

Dependent Variables:

Length of Time te Failure-Interval Open Date: This study measures 3
of the dependent variables, treatment retention, subsequent drug use, and
recidivism, not as dichotomous concepts but rather as a spread of time or
duration effect. In other words, treatment retention is measured as the
length of time or number of week an offender stayed in treatment as
opposed to either completed treatment or dropped out of treatment.
Similarly, this study is interested in the length of time an offender remains
“arrest free” or “drug free” versus rearrested or not rearrested, used drug or

didn’t use drugs. One way of controlling for differences in the length of

69

follow-up time without discarding data is by using survival analysis. This
statistical technique yields a survival function representing the conditional
probability of remaining in a given state over time; or in this instance,
“length of time to subsequent drug use”or “length of time to rearrest.”
Moreover, it provides information about patterns and the timing of an event
or condition (Pearson and Harper, 1990).

Three assumptions are made when using a survival model. First, for
any two individuals at any point in time, the ratio of their “risk of failure”
(hazard) is assumed to be constant over time (Motulsky, 1995). To be at
“risk of failure” assumes that the individual must be “exposed to the risk” or
in a condition where failure is a possible event. For example, an offender
cannot be considered to be at risk for dropping out of treatment, if in fact,
he is not in treatment. Second, a case ceases to remain “exposed to the
risk” of failure after termination of a condition (e.g., withdrawal from or
completion of treatment, rearrest, failed drug-screen); and third, “censored
observations” are treated as non-terminal “withdrawals”; the probationer is
not considered to have “failed or survived” (Petersilia and Turner, 1992).
Censored cases are simply dropped from the analysis at that point in time.
For this evaluation, the events which serve to “censor” a case from any of
the analyses include: death; lost to follow-up (no reported information on

the probationer for more than one month); or in the even supervision of a

70

probationer is transferred to anotherjurisdiction. In addition, some
offenders will not have “failed” at any event of interest (recidivism,
subsequent drug use) by the end of the follow-up period. Simply, the
observation period has not extended a sufficient period of time to capture
these possible events. These cases are also treated as censored data at
the point the offender was last observed.

Logically, this model assumes a starting point and end point, a
specified interval. Variables were created to specify the span of time each
offender was exposed to the risk of failing. Operationally defined, the
“Interval Open Date” variable represents the point in time when the
offender becomes “exposed to the risk” of failing. However, in this study
some offenders spent a period of time incarcerated after sentencing before
being released on probation. It cannot be assumed that the risk of failure is
constant during both the period of time an offender is incarcerated and the
period of time the offender is in the community. For those probationers in
jail or boot camp, the risk of failure (rearrest and failed drug screen) is
assumed to be extremely low, if it exists at all. In these cases the date the
interval opens was adjusted to the date where the ratio of the “risk of
failure” (hazard) was assumed to be constant. For example, if an offender
was sentenced to a term in jail and a term of probation, the “Interval Open

Date” is represented as the date in which the offender was released from

71

jail and began his probationary period. Similarly, if an offender was
sentenced to Boot Camp and a term of probation, the “Interval Open Date”
is the date the offender was released from Boot Camp and thus, becomes
exposed to risk of recidivism and subsequent drug use. For all other forms
of probation, the “Interval Open Date” represents the sentencing date.

Interval End Date: The interval end date is the date an offender was
removed from the “risk of failure”. This condition occurred when: the
offender was successfully discharged from probation; removed from
community placement via court sanction (e.g. jail or prison) for more than
60 days as a consequence of either a technical violation arrest or an arrest
for a new crime; or was still successfully on probation at the time the file
was reviewed. If an offender absconded and was not arrested at the time
the file was reviewed, then the warrant date was used as the “Interval End
Date”. A separate variable was created and coded to identify the end point
as: continued on probation; discharged from probation successfully; or
failure due to additional court sanctions. This variable was used to censor
the event in the Cox Regression models where recidivism is the dependent
variable.

Recidivism/Beeneet: Relying on official arrest records has
limitations. It is known that most crimes are hidden from enforcement

agents and are unlikely to be recorded (Tittle, 1980). Moreover, an arrest

72

does not necessarily mean a defendant is guilty. Lacking any other
measure of subsequent criminal behavior, this study did rely on officially
recorded arrest data in the probationers file. The rearrest variable
represents the date an offender was arrested for either a technical violation
or a new felony crime. An additional variable (Subsequent Violation Type)
was created to identify the rearrest as either an arrest due to a probation
violation or an arrest due to the commission of a new crime. Rearrests for
new crimes were also coded by type: property, drug, or violent. The
rearrest date did not necessitate a “failure” because an arrest for a
technical violation may not lead to a jail term of more than 60 days as
described above.

Subsequent Drug Use: All random drug screens required of
probationers are obtained at the adult probation department. Offenders
participating in residential probation are tested at the residential site while
in residence. Results are recorded in separate drug screen files at both the
adult probation department and the residential probation program. These
files were reviewed and both positive and negative results were recorded
for each probationer in the evaluation data set. The date a positive drug
screen was noted is considered the “failure” date. The time interval
between the “open interval date” variable and the drug failure date was

calculated to determine the length of time offenders remained drug free. In

73

the event an offender never tested positive for drugs during the follow-up
period, the length of time the offender remained drug free was calculated
from the “interval open date” to the “interval close date”. Because it was
determined that drug relapse for cocaine and other drugs are of greater
concern among officials in the study jurisdiction, and therefore court
sanctions are more likely to occur for subsequent cocaine/ other drug use
compared to relapse for alcohol and THC, it was important to analyzes

relapse for alcohol and THC use separately from relapse for cocaine and

 

other drugs. Thus, drug screen results for alcohol and THC were recorded
separately from cocaine and other drug screens. The major drug
categories screened for were: alcohol, THC, cocaine, and other. During
the evaluation period, drug screens were required on the average of once a
month. Probationers not required to submit to drug screens for more than
60 days were censored from the analysis at the date of their last drug
screen.

Treatment Access: All offenders included in the study sample were
court ordered to substance abuse treatment. This variable identifies
whether offenders complied with court mandated treatment or not and is a

dichotomous measure of the event.

74

Time in Substance Abuse Treatment:12 As previously mentioned,

offenders participating in residential probation were also required to
participated in out-patient substance abuse treatment. To hold constant the
modality of substance abuse treatment, only offenders sentenced to
probation and ordered to out-patient substance abuse treatment were
considered in the analysis. The “time in treatment” variable is therefore a
measurement of out-patient substance abuse treatment only. Unlike
residential treatment, which is measured in terms of treatment days, out-
patient treatment typically occurs on a weekly basis. Thus, for purposes of
this evaluation, “time in treatment” is measured in terms of “weeks in
treatment”. Information on the number of weeks between the treatment
admission date and the treatment discharge date was obtained from either
the probation files or was furnished by the treatment provider.

Treatment Status: Treatment status indicates the treatment
disposition of the offender at the time the offender was discharged from
treatment. The probationer was considered as either: never entering
treatment; unsuccessful at completing treatment; successful at completed

treatment; or successful and transferred for additional treatment.

 

12

“Time in Treatment” is also considered as an independent variable in the Cox regression
models where “time to rearrest” and “time to subsequent drug” are the dependent variables.

75

"thread;

Method of Analysis:

The statistical methods used in the research included univariant,
bivariant, and multivariant methods. To begin, summary statistics,
frequency, means, median and standard error measures were used to
ascertain and describe the basic profile of the sample and each cohort for
all independent variables. For comparing categorical data between
cohorts, Chi-Squares, Proportional z Tests, Fishers Exact test, Odds
Ratios and crude Relative Risk ratio (RR)13 methods are used. ANOVA
and t-Tests are used to test and describe differences among continuous
variables. These levels of analyses were necessary to understand the
degree of homogeneity within the sample and between the cohort
populations and, the complex relationships of the multivariant analysis.

Next, univariant survival analyses using SAS Life time test function
(SAS Institute Inc., 1997 pp. 413-431) were conducted for each time
sensitive outcome measure to illustrate differences in recidivism, treatment
retention and drug relapse patterns between the cohorts. As previously

discussed, survival analysis was used because the data represents time

 

The relative risk (RR) is used to measure the strength of associations found in
observational studies and is calculated from observed incidences rates of an event between two
cohorts. (From a 2x2 table; RR = [A/(A+B)]/ [C/(C+D)]). The greater the RR, the stronger the
association. A RR of more than 3.0 or more indicates a strong association, 2.0 indicates a
moderate association and a RR between 1.0 and 1.5 indicates a weak association. The RR for
observed incidence rates of events associated with study participants at the beginning of the
observational period (odds ratio) can be calculated from a 2x2 table where RR = (AxD)/(BxC)
(Lilienfeld and Stolley, 1994).

76

sensitive and censored data14 and because this type of analysis is
consistent with the substance abuse treatment and relapse literature . For
example, some probationers remained “crime or drug free” for one week
while others were not rearrest or experienced substance abuse relapsed by
the end of the observation period. Similarly, some probationers remained
in treatment only one week, while others in theory could have still been in
treatment at the end of the follow-up period. Both univariant and

multivariant survival analysis statistical techniques provide methods to

 

control for differences in the varying amounts of time individuals were
observed and whether or not the outcomes, or the events of interest,
occurred. The limitation to univariant survival analysis is that explanatory
variables are not controlled for. Thus, the analysis proceeds with
multivariant analyses including Cox regression, and logistic regression.
Cox regression, a mulitvariant survival analysis tool commonly
referred to as a proportional hazards regression (Motulsky, 1995; Kay,
1977; Cox, 1972) was then applied to evaluate the likelihood or the relative

risk of the outcomes examined: recidivism and drug relapse. This

 

14

Again, censored observations are treated as non-terminal “withdrawals”; the probationer
is not considered to have “failed or survived”. For this evaluation, the events which serve to
“censor” an individual from any of the analyses include: death, lost to follow-up (no reported
information on the probationer for more than one month), or in the event supervision of a
probationer is transferred to anotherjurisdiction. In addition, the event of interest may not have
occurred during the follow up period but may have occurred in the future. Simply, the
observation period was not a sufficient period of time to capture these possible events. These
cases are also treated as censored data at the point the offender was last observed.

77

statistical technique was believed to be the most appropriate in that the
method allows for the probability of the outcomes to be measures in
consideration of the varying amounts of time individuals were observed and
it controls for the various demographic and system factors which may
confound the relationship (Cox, 1972). This technique assumes that the risk
of failure between any two individuals at any point in time is constant and
uses regression methods to predict the relative risk of failing (hazard ratio)

based on one or more explanatory variables (Motulsky, 1995). For this

 

reason, Cox regression was not used to measure differences in the length
of time in treatment.

After examining the data it was established that the probability of
being determined a success or failure in completing treatment was
considerably different among offenders. For example, offenders sentenced
to residential probation were expected to complete 12 weeks of treatment
before they could be determined “successful”. In contrast, the length of
time offenders sentenced to other probation models were expected to
remain in treatment before being determined “successful” varied or was
unknown. Some offenders were considered successful after only one
month of treatment while others were noted as failures after six months of
treatment. Considering probationers sentenced to residential probation,

these offenders could not be successful in completing treatment during any

78

time period prior to 12 weeks. Thus at any point in time prior to 12 weeks
they had zero probability of success, and therefore a greater probability of
failure when compared to offenders with lower treatment expectations. In
other words, the probability of failing in treatment for offenders with longer
expected lengths of time in treatment is considerably different at earlier
points in time compared to offenders with shorter treatment expectations.
This also makes the interpretation of “failure” difficult. Failure for an
offender with a one month treatment expectation is significantly different
from failure for an offender with a 12 month treatment expectation. The
data indicate that the hazard functions of offenders sentenced to the
various probation models cannot be expected to be proportional over time.
In this situation, proportional hazard regression is not appropriate (Cox,
1972; Motulsky, 1995). Instead, because differences in the lengths of time
in treatment between cohorts is a continuous variable without any censored
observations, ANOVA statistics are used. Logistic regression statistics
were used to test differences in the proportion of offenders who actually
entered treatment. Like Cox regression methods, logistic regression
quantifies the association between risk factors and an event after adjusting
for other variables as relative risk ratio and require that the outcome
variables to be binary or dichotomous measures. However, in contrast,

logistic regression is appropriate when outcome variables are not time

79

sensitive.

In summary, several statistical techniques are used in this study. For
descriptive purposes summary statistics, frequency, means, median and
standard error were used. For comparing categorical and continuous
variables between cohorts, Chi-Square, Proportional z Tests, Fishers Exact
test, Odds Ratio, crude Relative Risk ratio, AVOVA and t-Test methods are
applied. Mulitvariant tools are employed for testing more complex

interrelationships among the data elements.

80

V. RESULTS AND FINDINGS

This chapter provides the results of the univarient, bivarient, and
multivariant analyses. The analyses consider three dimensions, the
probability of relationships, the strength of relationships and the direction of
relationships. One limitation in analyzing small samples is the potential
inability to detect significance for weak or moderate relationships. In
analyzing small samples it is therefore helpful to examine not just the
probability of significance, but also the strength and the direction of any
observed relationship.

This chapter begins with a comprehensive description of the
independent variables relating to the study sample and each cohort is
provided. These include demographic characteristics and the criminal
justice and substance abuse characteristics at the time offenders were
sentenced in 1996. Next, post-sentence criminal justice and substance
abuse offender characteristics and patterns of recidivism and treatment
retention are described. Lastly, results of the mulitvariant analyses and

ANOVA statistics employed to test the hypotheses are discussed.

81

 

Descriptive Analysis:
foender Sharacteristics at the Time of Sentencing:

The study sample consists of 209 drug involved felony offenders
sentenced in 1996 to a term of probation in a large midwestem county.
Because the intent of this evaluation study is to examine the effects of
residential probation on recidivism and drug relapse, each offender in the
sample also met the eligibility criteria for residential probation (The

rationale for the sample selection criteria were discussed previously in

 

Chapter IV). These criteria include: males with a minimum age of 17
years, who were convicted of a felony crime with sentencing guidelines of
“0 to 9 months" or greater or, who were convicted of a probation violation
on a felony with sentencing guidelines of “0 to 6 months".

Of the 209 probationers, 70 (33.5%) were sentenced to standard
probation, 54 (25.8%) to tether (PES,) 16 (7.7%) to boot camp with tether,
and 69 (33%) were sentenced to residential probation (SPCC). Of the 69
probationers sentenced to residential probation, 31 also received an
additional sanction which required probationers to be placed on tether
following discharge from residential probation. There is no question that
the experience for probationers sentenced to this additional level of
monitoring was different than the experience of probationers sentenced to

residential probation only. Therefore, the group of offenders sentenced to

82

residential probation in combination with tether was considered a as

separate cohort.

Sociegemegraphic Sharacterietics: The mean age for the total

sample of probationers is 27.9 years. Between cohorts the mean age
ranges from 21.4 to 30.7 years of age. Analysis of variance statistics were
applied to determine if observed differences were significant. As noted in

Table 2, the ANOVA P—Value is significant (.002); therefore t-Tests were

 

then used to determine significance between cohorts. Results show that
offenders sentenced to standard probation are considerably older than
offenders sentenced to Boot Camp/PES, SPCC and SPCC/PES.

Not surprisingly, a greater proportion 59.3% of the total population is
non-white than white. A marginal difference is noted between probationers
sentenced to residential probation and tether; 31.6% white vs 46.2% white,
but overall comparisons of white to non-white between cohorts are not
significant. With respect to education level, offenders sentenced to
standard probation were somewhat more likely to have achieved higher
education levels compared to offenders sentenced to Boot Camp/PES and

offenders sentenced to residential probation only (SPCC)”. However, it is

 

15

Due to the size of the cohorts, education level was collapsed into two categories (0-11
and 12 and above) for analysis.

83

.mmEoomm s 8888 8. v m as. 8% s 8888 8. v m . 585.335

.mmntasao 38m 3 889:8 2o. vm ”00mm 8 835288 mo. v m .flmbfla Ho.“ 8:588 895 +2 28 25mm 8.50360 .. "—33 553.6%

8:3on Eggm 9. 8.3988 5. v AT...

6:28 EN 8 8.89:8 S. v m _._ no?

888$ 885m u 8.5.8 ”85088888 u £38. 858 n mam 85088888 3:888 u 8.50% 88485 33:88.42 n 00%

 

 

 

 

 

 

g of. o msm o2 m.~m mm 03 2 0.2 «v cam 8>
m: own S 08 Z 55 am 5mm cm :6 am v.3 oz
EoE>o=—Em
um Q2 2 No 2.2 2 2: o 0.9 2 52 +2
am New m ”2 2 03 M: m.mm 3 2.2. cm 4.3 2350
mm m9“ 2 own 2 “saw cm 33 mm asm MN on 33me 2 - o
:33 :ozmoaum
v2 mam 2 own 2 GE 2 MS om two hm 5:2 333-82
mm Row M 9mm .2 fiwm mm New .2 cam m 2.24 8.33
z .x. z ..\o z .x. z .x. z ..\o Z .x. 00am
.56...
NA 4.2 NA m2 m2 Mm
mm 2. No 9.. mm 558934
2 2 M: 2 2 52552
m2 Own owm men m.wm 8:82
«EN 5.me Sm $3.8 mom :82
~88 m3
2.352 mmdBm meoonfi mun. comm not. .Bm
$524

mofimta> oEQEmoEouo_oom

 

mozmtouofiuzo oEquoEeuoBOw

N 039—.

84

reasonable to assume that the difference detected between offenders
sentenced to Boot Camp/PES compared to standard probation can be
explained by the age variance between the two populations. Finally, of the
total sample, 55% of probationers reported being unemployed at the time
of sentencing, with the greatest proportion of unemployed probationers
among the residential probation populations (SPCC = 68.4% and
SPCC/PES = 67.7%). Offenders sentenced to standard probation were 3

times more likely‘e’ to be employed at the time of sentencing compared to

 

both residential probationers and residential probationers with PES

combined.

Sriminal Justice and Subetence Abuse Ch4aracteristies: Table 3
depicts the criminal justice and substance abuse characteristics of the
study sample at the time of sentencing. In the total study sample, the
mean for all prior felonies is .51. Interestingly, the largest prior felony
mean (.60) is noted in the standard probation cohort, although no
significant difference in the number of prior felonies was detected between
the study cohorts (ANOVA; p= .28). However, differences in the nature of

the offence were noted (technical vs non-technical). Offenders sentenced

 

16

Calculated crude Odds Ratio which measures the strength of the association, as
previously discussed, from a 2x2 table where RR = {AD}/{BC} ( Lilienfeld and Stolley, 1994).

85

002m 8 85888 mo. vm was mmfioomm B8 $5 3 8.89:8 5. VA .. 880 88.2.20
888 85o :u 3 8888 EVA 4 “888 8822:

48485 288m u 48.4.8 854854058 n 8.5m 8:8 n ma.— 85828485 8828 n 8.86% 888on 8888 u 8%

 

 

 

 

 

 

mEocoEow um mozmtouoflmco om3n< oocmuwnzm new 832:. .mEEtO

Ne 4.44 m 82 N_ 24 M: 2m 2 4.44 _N 8m wagon
Nm m: N mi: N no 2 D: m N: N_ H: 8453588
3 84 N com N_ is 4N 4.44 4_ 8m AN 84 05
8 4.4 N as 4 as N Z N 3 2 n: 88?.
86:0 No 9:0
mm NNN m 34. N_ is 2 Nam 2 N.4m 4 3.2 89>
S on 4 :4. N2 53 2 4.2 : 3N NN 4.: 9485
44 N._4 m a: N SN 2 N? 42 Nam 3 63 man
8385820 88:0
42 4R 2 N.; N :4 N4 NR 4N NS 8 44.44 8.88882
44 EN 4. 4:: 2 NNN N_ NNN 42 New 4 3 28.889
2 .x. z .x. z .x. z .x. z .x. z .x. 8;» 8:80 88.26:.
.35
2.0 NS NS 2.0 :8 mm
2 m m m 4 8:882
o o o o o 8:582
o o o o o 4282
43 N8 So who 8.0 :82
de 8:63“. .2...
u:_u>.n_ 898 8.808 mum comm 85 .65
<>02<

mo_n~_._m> omzn< mocmumnzw use 833.. 35.55

m 23¢...

86

to standard probation were less likely to be technical violators.” Only
5.7% of offenders in the standard probation cohort were sentenced due to
technical violations compared to 63.2% of the offenders in SPCC, 67.7% in
SPCC/PES, 77.8% in PES and 81.2% in BC/PES. In other words,
offenders sentenced to SPCC, SPCC/PES, PES and BC/PES combined
were 6.4 times more likely to have been sentenced as a result of a
technical violation compared to offenders sentenced to standard probation.
Regarding offence classification, the largest proportion (41.2%) of
offenders in the sample were convicted for drug offences, followed by
property and violent offences (32.0% and 27.8% respectively).18 In
comparing offence classifications between cohorts, offenders sentenced to
standard probation represented the largest proportion of drug offences
(55.7%) and the smallest proportion of violent offences (12.9%). Offenders
within the standard probation cohort were over 3 times more likely19 to be
sentenced as a result of non-violent crimes (drugs and property) compared

to all other cohorts combined. Sentences for violent crimes represent

 

l7

Chi-Square statistics were used to test overall significance (p<.01). However, due to the
size of some of the cells, Fisher Exact test statistics were employed to determine significance
between cohorts.

18

Both technical violations and non-technical violations are included. For offenders
sentenced on a technical violation, the classification of the original offence was considered.

19

Drug and Property categories were collapsed to represent the Non-Violent category.
Chi-Square p < .05. Odds Ratio calculation = 3.6.

87

34.2% of all offences among offenders sentenced to SPCC, 38.7% of the
offences within the SPCC/PES cohort, and 35.2% and 31.3% of the
offences among PES and BC/PES offenders respectively, compared to
only 12.9% of all offences within the standard probation cohort.

The most frequently reported drug of choice in the total sample is
marijuana (40.7%) followed by poly-drug use20 (34.4%). Due to the size of
the cohorts, the drug categories were collapsed before differences between

cohorts could be tested“. The comparative analyses show that drug of

 

choice is distribution between cohorts consistently with no significant

differences noted.

Qriminal Justice Qharageristics: Post gentencing:
Actual Jail Days: Many probationers in the study sample received a

split sentenced of jail terms in addition to probation. Although not
empirically documented in the literature, it seemed reasonable to assume

that the split sentence in and of itself would impact offender outcomes.

 

20

This category represents offenders who reported more than one drug of choice or the
use of multiple drugs in the same classification e.g., barbiturates; central nervous system
depressants etc.

21

The comparative analysis was performed in two phases. First, three categories were
created by collapsing Alcohol and THC into one category. Chi-Square statistics were performed,
with no significance noted. However, the validity between two of the cohorts remained
questionable. The drug of choice was further collapsed into two categories (Alcohol/THC and
Cocaine/Crack/Poly-drug). The rational for this bifurcation was previously discussed. Following,
both a standard Chi-Square and Fisher Exact tests were performed with no significant
differences noted.

88

Therefore, the use of a split sentence was documented and included in the
mulitvariant analyses as an independent variable. The variable “Jail Days”
represent the length of time each offender spent in jail after the critical
incident arrest and any time served after receiving a split sentenced. Table
4 illustrates the differences in the average length of time offenders in
different cohorts actually spent incarcerated. The mean jail time ranged
from 38.4 to 101.1 days. The overall comparison is significant (ANOVA P-

Value = .005). The t-Test was used to test for differences between each

 

cohort. These tests demonstrate that there are significance differences
between offenders sentenced to standard probation and the other cohorts
in the length of time offenders are incarcerated ranging from p < .01 - p <
.05. Comparisons between SPCC, SPCC/PES, PES and BC/PES cohorts

are not significant.

Sentence Type: The finding that probationers sentenced to
standard probation spend less time in jail is interesting in view of the
proportion of offenders who received split probation sentences and the
distribution of this sentences between cohorts. Of the total sample, 40.7%
of all offenders received split sentences. When the proportions of split

sentences for each cohort are examined and comparisons made between

89

88858 0388 83:28 8 0885?. Hon mum 298m .. ”.8335820 coach—O mac—oh «nos—oomnsm

8&de 9 woamfioo S. v m: 505.9% 8 88988 mo. v m g “as noun—o5 32:53.5

.mmncum can mmE .mmndoomm 3 @8368 S. vm 98 00mm 8 839:8 mo. V9. “995 :3. 3504.4

8888 88:8 n sen—.48 8885688 n 8&8 85$ n ma.— ..saoeaoesoa 3822QO u 8.50% .8888 38888 u 008

Q:

 

 

 

 

 

o o fl m.mm m OR o o m cam 3:205
4m Q? m 2: N 8.8 m 0N8 S 2: S 8.8 3885-82
2282.885 88:0 >58“. «eosdmmnzw
m4 QON m 3 m m 3 m mi 2 SN 2 m.4N 820m 82
4m New w «9% E 44w+m mm 4084 S «:44 2 04m 8:302.
mm Nam m MK: 2 m.mm S 3% Z odm om «Am 282
2;... 4.05205 “cosuomnzm
4S mom 2 m8 om n40 4m o? 2 m8 24 o? 8:305
mm hoe o msm : 0mm om osm 2 ~.4m 8 08. £39 8:805:84
z .x. z .x. z .x. z .x. z .x. z .x. cm? 8523
.85.
NS 0.2 mi 8.2 g Mm
Sm w0m mom 9% M: m 82582
o o o N o 5:882
new cam 93 24 oN 8.82
9mm 3% :2 WE .44an 582
33 «>3 :2. .8844
u:_u>.n_ oomflom mmaaoaw wma 00am sci .Em
<>Oz<

m:_o:3cow amon— Hmozmtoaofiuco oozmz... .mEEtO

832...; 8:2... .8845

4 28¢

90

cohorts, very small and insignificant differences are noted. Among all five
cohorts, the maximum difference in the proportion of split sentences is only
7.5%, ranging from 30.0% to 37.5%. One might conclude that although the
relative risk of receiving a split probation sentence is approximately the
same across probation models, the actual jail time offenders receive is

significantly less for offenders sentenced to standard probation.

Subsequent Violation Type: One strength of this research design is
that it was possible to track offenders for a minimum of 24 months. For
offenders sentenced in early 1996, the follow-up period extend up to 34
months. This time period is believed to be adequate to test the effects of
residential probation with treatment on recidivism. As previously
mentioned, policy makers supporting this evaluation are particularly
interested in the commission of new felony crimes as a measurement of
recidivism. However, in many cases offenders are often faced with severe
consequences for technical violations and are removed from community
supervision for a considerable length of time. This presents limitations on
measuring subsequent felony arrests when survival analysis methods are
used. Survival analysis assumes a well-defined end point and recurring
events should not be analyzed with this methodology (Motulsky, 1995).

When offenders are rearrested and receive additional court sanctions that

91

remove them from community supervision, they have by definition failed,
even though they may return to community supervision after 60 days or so.
In a survival analysis, this failure is treated as a terminal end point and the
case is not included in the accumulative proportion still surviving at the next
time interval. Cases can not be reentered in the analysis at a later time to
measure a reoccurring event, or in this case a rearrest for a new felony
following reentry into the community. It is therefore not possible to follow

offenders who have failed due to a technical violation past the technical

 

violation failure date to determine if a future felony crime was committed.
Thus, it is recognized that the measure of recidivism due to new felony
crimes within the follow-up period may be an underestimate.

In the total study sample, 39.2% of offenders had not been rearrest
and court sanctioned (failed)22 by the end of the follow-up period. Figure 1
and Table 5 provide a visual representation and the descriptive statistics of
the survival rates for all court sanctioned failure including both technical
violations and new felony crimes for each cohort.23 For all survival analysis

in this evaluation, observations are censored in the event of death, if the

 

22

This distinction is important to the analyses. A rearrest was not considered to be synonymous
with failure. Again, failure was operationally defined as a rearrested that resulted in a court sanction
which removed the offender from community supervision for more than 60 days. In cases where offenders
were rearrested but were not considered to have failed, the offender was either rearrested and the case
dropped or rearrested and immediately placed back into the community and continued on probation.

23

All univariant survival analysis graphs and associated descriptive statistics were
constructed using the SAS Lifetime Test function (SAS Institute Inc., 1997).

92

35.5 >53... 52 as «5.635 E223... 8:828 .58 .2863 2.238 9.8;

$an 0 mumfiomm o mum 4 00mm 0
fix «a 9.33

E: a: of 3: cm ow 3

F F b _ _ _ k

 

 

'
¢

 

 

5:32“. .30. *

om

 

 

 

 

 

 

 

 

P use“.

2
.3
.2
.3
-3
-2
.2
-S
.3

-3

 

-03

Buumuns «y.

93

Table 5 Weeks Surviving Additional Court Sanction: Technical
Violations and New Felony Crimes

 

A = Std. Prob B = SPCC C = PES D = SPCC/PES E = BC/PES
Mean 91.8 62.3 75.5 65.4 47.7
Median 110.4 49.4 73.9 46.0 25.2
SE 7.1 8.3 7.8 8.4 11.1

 

offender is lost to follow-up or if supervision is transferred to another
jurisdiction.

When technical violations and new felony crimes are examined
separately, 40.2% of all offenders failed due to technical violations and
20.6% of the offenders committed a new felony crime. Between cohorts,
significant differences were noted when comparing offenders sentenced to
standard probation and each cohort. Offenders sentenced to standard
probation were less likely to be rearrested and court sanctioned for
technical violations compared to offenders sentenced to each of the other
probation models. This is consistent with previous research which show
that rearrest rates for more intensive probation programs are higher.
Specifically, when monitoring procedures are intensified, the likelihood that
a probation violation will be identified is greater (Petersilia and Turner,
1993b). Figures 2 and 3 illustrates the survival analyses for technical
violations and new felony crimes respectively. As in Figure 1, the survival
analysis and descriptive statistics in Figure 2 and Table 6 includes all

94

 

«ceases 5.5.8.. 2.225 9.33 N 2:9“.

mmEOm o mmmaomm . mum 4 00mm 0 5:235 .36 *
is an 8.33

of a: of 2: cm 2 3 E c

2 _ _ _ _ _ _ _ 2

g
L:
-3

Z:

_ -3

 

 

 

 

r—dI
<>

 

 

 

 

 

 

 

95

fiumuuns %

Table 6 Weeks Surviving Technical Violations

 

 

A = Std. Prob B = SPCC C = PES D = SPCC/PES E = BC/PES
Mean 115.6 78.4 87.2 73.6 67.8
Median * 91.6 91.9 68.6 69.7
SE 6.7 9.5 8.4 8.8 13.6

More than 50% still surviving at the end of the follow-up period

 

offenders. However, the single event that establishes failure in this model
is obviously a court sanctioned technical violation that removed an offender
from community supervision for more than 60 days. Offenders who are
arrested for new felony crimes prior to an observed technical violation are
censored from the analysis at the time of the arrest. The rationale to this is
straight forward. All offenders, including those who are arrested for new
felony crimes, are at equal risk for a technical violation failure at the
beginning of each offenders observation period. Offenders who are
arrested for a new felony crime, but who did not experience a technical
violation failure up to the time of the arrest, by definition have not failed in
this analysis. However, because an arrest for a new felony crime removes
the offender from community supervision (jailed or sent to prison), these
offenders are no longer at risk for failing the event of interest and therefore
the case becomes a censored event. Including all offenders in the model
and censoring on new felony crimes provides a more accurate analysis of

technical violation rates.

96

Figure 3 and Table 7 represents the survival analysis for new felony
crimes. The same procedures as described for Figure 2 were employed
with the exception of the censored events. Since the event of interest is
new felony crimes, the censored events in this analysis include technical
violation failure as well as those previously mentioned (death, lost to

follow-up, supervision transferred to another jurisdiction).

Subsequent Eelony Offence classification: As noted above, by the

end of the observation period 20.6% of all offenders in the study sample

 

had failed due to the arrest and conviction of new felony crimes. It was the
intent of this study to measure whether there were differences in the
severity of the subsequent felony crime (e.g., violent vs. non-violent)
between cohorts. However, based on results of the descriptive statistics,
this question can not be answered. Overall, only 9 (21.0%) violent new
felony crimes and 34 (79.0%) non-violent felony crimes were recorded
among the study sample. When distributed among the five cohorts, the
incident rate within each cohort is not large enough to make any stable

estimates.

97

35.5 .320... :62 2.225 9.83 n 2:2...

mun—Em o mun—60mm a mum 4 00mm 0 5:32“. .Bm *
.3... .a 9.33
of o: o: E: on ow 3 S o

. _ L F L _ _ b _ _

A:
.3

L:
e

 

.3

 

 

 

-3

.3
H E I...
a . -3
4.4 AA“! .. 3
$9! .2:

 

 

 

 

r—-)L
(-

 

 

 

 

 

 

 

98

Buuwuns 0/.

Table 7 Weeks Surviving New Felony Crimes

 

 

A = Std. Prob B = SPCC C = PES D = SPCC/PES E = BC/PES
Mean 110.0 86.0 114.3 103.6 60.8
Median * * * * 58.7
SE 7.6 11.4 9.8 14.3 18.1

More than 50% still surviving at the end of the follow-up period

 

Substance Abuse Qharacteristics: Post Sentencing:
Received Substance Abuse Treatment: Previous research has

consistently demonstrated that the “length of time” a client remains in
treatment is the most constant and best predictor of treatment outcomes.
This is true even among clients who drop out of treatment (Simpson and
Friend, 1988; DeLeon, 1988, 1994; Hubbard,1988; Barr and Antes, 1981).
Moreover, studies support the conclusion that how a substance abuser is
exposed to treatment seems irrelevant. What is important is that
substance abusers are brought into an environment where treatment can
occur (Wexler, et al., 1990; Fields, 1989; Langenauer and Bowden, 1971;
Stephens and Cottrel, 1972; Anglin, 1988; De Leon: 1988, Anglin and Hser,
1990a; Lipton, 1995). Unfortunately only a small proportion of offenders in
need of treatment actually receive treatment. The effects of treatment can
only be measured if the offender actually enters treatment. It is therefore

important to determine if there are differences in the proportion of offenders

99

who enter treatment between probation modalities. Referring to Table 8,
only 58.3% of all offenders sentence to substance abuse treatment actually
received any treatment. When comparisons between cohorts are
examined, it is noted that offenders sentenced to residential probation
(SPCC) or residential probation with tether (SPCC/PES) are approximately
1.5 times more likely to enter substance abuse treatment compared to all
other cohorts“. Only 52.9% of offenders sentenced to standard probation
actually entered substance abuse treatment compared to 81.6% of SPCC
and 77.4% of SPCC/PES probationers. The least likely to enter substance

abuse treatment are offenders sentenced to Boot Camp/PES (31.2%).

W: Within the total sample, among

offenders who actually entered treatment, 77.9% successfully completed
treatment. Comparisons between cohorts were not significant. The
proportion of offenders successfully completing treatment is relatively
consistent, ranging from 71.0% to 86.5%. It must be noted that this
measurement is not an indicator of the length of time in treatment. For
example, as previously discussed the expected length of time in treatment

for offenders sentenced to either residential probation model

24 Overall Chi-Square; p < .01. Crude Relative Risk calculations; RR = 1.45-1.54

100

8:32.. 32% u £525

.995 .9m 8 3.388 S. vm « "Eon—«8:. 3.2:. 8533.6 3233—

..eeotgamosom u mice 858 n 5.. 853.86%... .6623 u 3.50% 92.35 .6583. u 66%

 

 

 

 

OR a... h... Q... We mm
0.; 5.9: Odo ©on owe 5:582
A: ON ON QM A: 6:8;
mw 0.2 fmm fim. CG 53602
5.3 QNN o.w~ «.43 odm 53:
de
33>... 2.253.... 5 9.3.5
<>OZ<
hm fimm _ odm o owm o oém o odm m We Emmooosmdb
mm afih v Wow w. own 2 0.3. mm 9:. mm mow Emmooozm
mafimmgasomfi 2253....
N ON C o.o fl N.m fl ON o od 0 od :BOSED
mm Wow 2 ”we 0 v.2 mm wAm h v.2 mm 554 . OZ
N2 m.wm m Ngm *VN v.21 mm New .Lm 0.5 mm QNm mo>
Z .x. Z o\.. z o\o Z c\o Z o\o z o\o «co—Snot. omsn<
.59.. oucSmnsw oozooom
mun—5m mun—BUn—w mum 00am 02m .Bw

mo_am_.m> omzo< oocflmnsw

mEocoucow “mom "moumtouoflmco omsn< oocflmnsw w 206...

l
0
1

(SPCC or SPCC/PES) is approximately 14 weeks. Once offenders
complete Phase One and Two of the residential probation out-patient
substance abuse treatment model, they are considered to have met the
treatment expectation and are discharge as “successful completers.”
Offenders sentenced to other probation models may enter out-patient
treatment delivered by an alternative treatment provider. The expected

length of time in treatment appears to vary from provider to provider”.

Weeks in Treatment: Time in treatment is considered as both an
independent and dependent variable in this evaluation study and is
discussed here because of the predicted relationship of this variable to the
other dependent variables. As noted, not all offenders sentenced to
substance abuse treatment actually enter treatment. However, of those
who entered treatment, the mean length of time in treatment is 23.2 weeks.
Overall comparisons between cohorts are not significant (AVOVA P-Value
= .33) ranging from 14.6 to 28.8 weeks. Figure 4 and Table 9 provide a
graphic illustration and the descriptive statistics respectively of the length
of time in treatment for each cohort. In this analysis there are no censored

observations. All offenders who entered treatment either completed

 

25

Treatment expectation between providers vary. The majority of offenders in SPCC and
SPCC/PES received substance abuse treatment from the same treatment provider. Overall, 3
treatment providers delivered serviced to the study cohort, and all services were based on an
out-patient treatment model.

102

.558... 322 8:53am 5 9.8;

a 2:9.

 

 

 

 

 

 

 

 

 

 

 

 

 

8&8 o 3.60% . mm. 4 8% o 8:22.. .2». ...
.5525. =_ 9.33
O: 2: 3 O. E 2 on 3 a S E
. H _ _ _ _ H _ _ _ _
a V... 2 b a
.. rlllllllw . 6
(law. .
-1
4.
J14 V

 

 

 

 

 

 

 

 

 

 

 

TE

.3

.3

.3

-3

A:

L;

.3

-2:

 

Buuwuns %

103

Table 9 Weeks in Substance Abuse Treatment

 

 

A = Std. Prob B = SPCC C = PES D = SPCC/PES E = BC/PES
Mean 28.5 18.5 28.8 22.9 14.6
Median 21.0 12.0 23.0 16.5 8.0
SE. 4.5 4.0 4.7 4.9 7.1

 

treatment or were discharged as unsuccessful by the end of the

observation period.

 

Descriptive Summary:

Results of the univarient and bivariant analyses suggest there are
only marginal differences relative to offender characteristics between
offenders sentenced to SPCC, SPCC/PES, PES, and Boot Camp/PES at
the time of sentencing. However, offender characteristics among
probationers sentenced to standard probation appear to vary on several
dimensions. Offenders sentenced to standard probation tend to be older,
have a higher level of education and are more likely to be employed. They
are also less likely to be technical violators at the time of sentencing and a
greater proportion of their crimes are drug and property crimes. Although
offenders in this cohort are just as likely to receive a split sentence

compared to the other cohorts, the actual time these offenders spend in jail

104

for the offence is less compared to all other cohorts”. Thus, the degree of
homogeneity between offenders sentenced to standard probation and
offenders sentenced to the other probation models is less. Where
appropriate, mulitvariant tools were used to control for these differences in

testing the main hypotheses.

Testing the Hypotheses:

This phase of the analysis specifically tests each hypothesis using
Cox regression, logistic regression and ANOVA statistics. The rationale for
the statistical methods and the procedures applied in constructing each
model used to tests the hypotheses are briefly restated to set the
framework in which each hypothesis is tested. Following, the results of
each analysis are presented.

Cox regression analysis was used to test differences in the outcome
measures of recidivism and subsequent drug use because the data
represents time sensitive and censored data. This technique allows for
both the probability of the outcomes to be measures in consideration of the

varying amounts of time individuals were observed and controls for the

 

26

As previously discussed the “Time in Jail" variable was calculated to include the days
each offender spend in jail for the offence from the date of arrest to sentencing, and any
additional time served on a split sentence. The actual length of the jail sentence is not known.
Therefore it can not be determined if the length of time in jail is a function of the jail sentence or
other factors such as the ability to make bond or early release.

105

various demographic and system factors which may confound the
relationship. Cox regression assumes that the risk of failure between any
two individuals at any point in time is constant and uses regression
methods to predict the relative risk of failing based on one or more
explanatory variables. For this reason, Cox regression was not used to
measure differences in the dependent variable “length of time in treatment”
even though this outcome variable is time dependent.

After examining the length of time in treatment data it was
established that the probability of being determined a success or failure in
completing treatment was considerably different between offenders.
Offenders sentenced to residential probation were expected to complete 12
weeks of treatment before they could be considered “successful”. In
contrast, the length of time that offenders who were sentenced to other
probation models were expected to remain in treatment varied both within
and between probation models and was as little as one month.
Probationers sentenced to residential probation had zero probability of
successfully completing treatment before 12 weeks, while others with
shorter treatment expectations had some probability of successfully
completing treatment prior to 12 weeks. Therefore the probability of failing
in treatment for offenders with longer expected lengths of time in treatment

is considerably different at earlier points in time compared to offenders with

106

shorter treatment expectations. This also makes the interpretation of
“failure” difficult. Failure for offenders with a one month treatment
expectation is significantly different from failure for offenders with a 12
month treatment expectation. These data indicate that the hazard functions
of offenders sentenced to the various probation models can not be
expected to be proportional over time. In this situation, proportional
hazards regression is not appropriate. Instead, because differences in the
lengths of time in treatment between cohorts is a continuous variable
without any censored observations, ANOVA statistics are used to test for
the significance of differences in the length of time offenders remained in
treatment between cohorts.

Like Cox regression methods, logistic regressions quantifies the
association between risk factors and an event as a relative risk ratio after
adjusting for other variables. However, this statistical method requires the
outcome variables to be binary or a dichotomous measures and is
appropriate when the outcome variables are not time sensitive. For these
reasons logistic regression statistics are used to test for the significance of
differences in the proportion of offenders who actually entered treatment.

Before preceding with the results, it is first necessary to describe the
procedures used in constructing the multivariant models. This will provided

ease in interpretation for the reader. In the Cox and logistic regressions

107

relative risk ratios for categorical variables are calculated in relation to a
reference variable or study group (referred to as the omitted category or
the hold out). Typically the reference study group is the cohort of interest
to which all other cohorts are compared. This study is focused on the
effects of residential probation on outcomes; thus residential probation is
the variable of interest. However, in this study all regression models use
standard probation as the reference group for two reasons. First, the
degree of homogeneity is much less compared to the other cohorts.
Second, residential probation is subdivided to create two separate
residential probation models. Using a reference group other than one of
the residential probation models makes it easier to visually examine the
risk ratios between the two residential probation models.

Hold outs (also referred to as reference or omitted category) for all
other independent categorical variables are also consistent in all the
models and are as follows. In the data set ethnicity is recorded as “Race”
and consists of two categories, white and non-white”. The non-white
category is always considered the hold out. Employment is represented as
a dichotomous variable, “ not employed or employed,” and “not employed”

is the omitted category. Drug of choice is collapsed into two categories,

 

27

The procedures and rationale for combining categories was previously discussed and
can be found in Chapter IV. Methods; Variable and Measures.

108

“alcohol/THC” and “cocaine/polydrug”. The category alcohol/THC
represents the reference category. The offence variable consists of three
offence types, “violent, property and drug The violent offence category
is selected as the hold out group. Probation violation indicates whether the
critical offence which led to the 1996 sentence was the result of a technical
violation or not. Here, the non-technical violation category is the omitted
category. The independent continuous variables were constructed in each
model to be interpreted in ascending order. For example, in the variable
age, the risk ratio is associated with older offenders relative to younger
offenders. Forjail days, the risk ratio is associated with more days in jail
relative to less days in jail, and for time in treatment, more time in
treatment relative less time in treatment.

Each model tested using Cox or logistic regression was performed in
two phases. First, all of the independent variables previously defined in
Chapter IV. were included in an initial analysis (full model). From this
analysis the independent variables which were identified as either
significant or approaching significance (p<.10) and the cohort variables
were entered into a second regression analysis (reduced model)”. This

second analysis was done for several reasons. By including only the

 

28

Because this analysis is intended to distinguish between the various cohorts, the cohort
variables are always included in the reduced analysis whether they are significantly related to the
dependent variable or not.

109

variables from the full model which are shown to be approaching
significance or are significant in a reduced model improves the power of
the model. In other words, this increases the probability that true
differences will be detected. For example, variables approaching
significance may reach significance when the power of the model is
improved. This process is particularly useful when analyzing a sample with
a limited number of observations. In addition, this procedure provides a
more clear assessment of the strength in the relationships between the
independent and dependent variables. Presenting a reduced model is the
most parsimonious way to provide a clear analysis of these multivariant

models.

Recidivism:

To begin, an overall recidivism failure model was constructed to
examine differences in the length of time from sentencing to additional
court sanctioned failure. Court sanction failure was defined as a
subsequent arrest which led to a court sanctioned that removed an offender
from community supervision for more than 60 consecutive days. This
analysis was only concerned with examining the relationships of the
independent variables on “overall failure” and did not make any distinction

between failure due to rearrests for new felony crimes or rearrests for

110

technical violations. Table 10a presents the first phase in this analysis
(refer back to the survival analysis plot, Figure 1, for visual illustration).
Offenders sentenced to either residential probation program (SPCC and
SPCC/PES) are over 2 times more likely to experience a court sanctioned
failure over time compared to standard probation while offenders
sentenced to tether (PES) and boot camp with tether (BC/PES) are 1.5
times more likely to fail over time. This supports the bivariant analysis
discussed and is consistent with previous research which show that
rearrest rates for more intensive probation programs are higher.
Specifically, when monitoring procedures are intensified, the likelihood that
a probation violation will be identified is greater (Petersilia and Turner,
1993b). Additionally, Table 10a shows that length of time in treatment,
offence category and drug of choice are related to failure. As might be
expected based of several studies, the single best predictor of failure in the
model is the length of time in substance abuse treatment (p<.0001).
Offenders who remain in treatment longer are less likely to fail. The offence
category is also shown to have a significant effect on length of time to
failure (p<.01). Offenders sentenced for drug crimes are 2 times more
likely to fail over time compared to offenders sentenced for violent crimes
(the reference category). Drug of choice is also noted to affect failure in

this model. Offenders who report their drug of choice to be cocaine/crack or

111

Table 10a

Length of Time Until Court Sanctioned Failure: Full Model

 

Summam of the Number of Event and Qansgreg Values

Percent
Total Event Censored Censored
201 122 79 39.30

Testin Iobal Null H othesis: BETA=0
Without With

Criterion Covariates Covariates

-2 LOGL 1161.826 1060.809

Score

Wald

Analysis of Maximum Likelihood Estimates

Parameter
Variable DF Estimate
Age 1 -0.018004
Race 1 -0.194632
Education Level 1 -0.062422
Employment 1 -0.099179
Prior Felonies 1 0.132237
Offence/Property 1 0.183087
Offence/Drug 1 0.724856
Jail Days 1 0.000599
Drug of Choice 1 0.498234
Treatment Weeks 1 -0.064471
Probation Violation 1 -0.157513
SPCC 1 0.884449
PES 1 0.434964
SPCC/PES 1 0.801024
BC/PES 1 0.432593
Linear Hypothasas Taating

Wald

Label Chi-Square
COHORT CATEGORY 8.9383
OFF ENCE CATEGORY 8.9048

Model Chi-Square

101.017 with 15 DF (p=0.0001)
74.199 with 15 DF (p=0.0001)
50.717 with 15 DF (p=0.0001)

Standard Pr > Risk

Error Chi-Square Ratio
0.01503 0.2310 0.982
0.21396 0.3630 0.823
0.22939 0.7855 0.939
0.20052 0.6209 0.906
0.14101 0.3484 1.141
0.28581 0.5218* 1.201
0.26434 0.0061* 2.064
0.00109 0.5845 1.001
0.22505 0.0268* 1 .646
0.01208 0.0001* 0.938
0.24360 0.5179 0.854
0.31882 0.0055* 2.422
0.27602 0.1151* 1.545
0.33067 0.0154* 2.228
0.34713 0.2127* 1.541

Pr >
OF Chi-Square
4 0.0627*
2 0.0117*

 

* Used in reduced model

112

polydrugs have significantly lower lengths of time to court sanctioned
failure.

In the second phase, the reduced model was created. The
independent variables, offence category, time in treatment and the cohort
category were then included in the analysis to better assess the strength of
the relationships between the independent variables observed as
significant or approaching significance in the full model and the dependent
variable. Table 10b specifies the results of this analysis (refer back to the
survival analysis plot, Fiqure 2, for visual illustration). As can be seen, the
relationships of the independent variables identified as significant
predictors of court sanctioned failure in Table 6a are strengthened in this
model. Again, weeks in treatment, offence category and drug of choice are
significantly related to the length of time offenders remain in the community
before failing and the relative risk of failing over time is greater for
offenders sentenced to residential probation.

Next, failure due to technical violations and new felony arrests were
examined separately. This was accomplished by running two separate
analyses and censoring the event that was not of interest. Specifically, in
the model that examines technical violations, new felony crimes are
censored events whereas in the model that examines new felony crimes,

technical violations are censored. First, technical violations are examined.

113

Table 10b Length of Time Until Court Sanctioned Failure: Reduced Model

 

mma of the Number v nta d
Percent

Total Event Censored Censored
201 122 79 39.30

Test'n lobal Null H othesi :BETA=0
Without With

Criterion Covariates Covariates

—2 LOGL 1161.826 1066.054

Score

Wald

Analysis of Maximum Likelihood Estimates

nsored V Iues

Model Chi-Square

95.771 with 8 DF (p=0.0001)
64.794 with 8 DF (p=0.0001)
56.106 with 8 DF (p=0.0001)

 

Parameter Standard Pr > Risk
Variable DF Estimate Error Chi-Square Ratio
Offence/Property 1 0.145672 0.26179 0.5779 1 .157
Offence/Drug 1 0.729965 0.23856 0.0022* 2.075
Drug of Choice 1 0.506820 0.19779 0.0104“ 1.660
Treatment Weeks 1 -0.067389 0.01 169 0.0001 * 0.935
SPCC 1 0.997845 0.29490 0.0007* 2.712
PES 1 0.555082 0.25981 0.0326“ 1 .742
SPCC/PES 1 0.875839 0.31007 0.0047“ 2.401
BC/PES 1 0.571 101 0.33578 0.0890 1.770
Linaar Hypotheses Testing

Wald Pr >
Label Chi-Square DF Chi-Square
COHORT CATEGORY 13.3614 4 0.0096
OF FENCE CATEGORY 11.5118 2 0.0032
*p<01

“p<05

H4

As in the previous model, all independent variables were initially introduced
into the analysis (Table 11a). Only those variables identified as significant
or approaching significance (p<.10) were included in a second reduced
model of analysis. Table 11b illustrates these results. Here again, the
single best predictor of positive outcomes is the length of time in substance
abuse treatment (p<.0001). The longer offenders stay in treatment the less
likely there are to be arrested and court sanctioned for technical violations.
In addition, the level of monitoring is directly related to technical violations.
Offenders sentenced to the most intensive probation model are the most
likely to fail. Specifically, offenders sentenced to residential probation with
tether are 3.6 times more likely to be arrested and court sanctioned for
technical violations compared to the least monitored cohort, standard
probation. The next most intensely monitored probation model is
residential probation. As expected, these offenders are the next most likely
to be arrested and court sanctioned for technical violations followed by
tether and boot camp: 3.1, 2.5, and 2.1 respectively. These data strongly
supports hypothesis 3a which states, “the risk of failure over time for
technical violations is greater among offenders sentenced to residential
probation” (refer to section IV Methods).

Of particular interested in this study was the commission of new

felony crimes as a measurement of recidivism. However, as shown above

115

Table 11a
Full Model

Length of Time Until Court Sanctioned Technical Violations:

 

Summary of tha Numbar sf Event and Censored Values

Percent
Total Event Censored Censored
201 81 120 59.70
Testin lobal Null H othesis BETA=0

Without With

Criterion Covariates Covariates
-2 LOG L 774.251 704.092
Score
Wald

Analysis of Maximum Likelihood Estimates

Parameter

Variable DF Estimate
Age 1 0029964
Race 1 -0.219918
Education Level 1 0.316298
Employment 1 -0.231399
Prior Felonies 1 0.043835
Offence/Property 1 0.198429
Offence/Drug 1 0.549576
Jail Days 1 0.001068
Drug of Choice 1 0.350501
Treatment Weeks 1 —0.060761
Probation Violation 1 -0.287498
SPCC 1 1 .26101 1
PES 1 0.908618
SPCC/PES 1 1.301721
BC/PES 1 0.674566
Linear Hypathesas Testing

Wald
Label Chi-Square
COHORT CATEGORY 12.3476
OFFENCE CATEGORY 3.2858

Model Chi-Square

70.159 with 15 DF (p=0.0001)
50.694 with 15 DF (p=0.0001)
42.396 with 15 DF (p=0.0002)

Standard Pr >
Error Chi-Square
0.01995 0.1330
0.26039 0.3983
0.28660 0.2698
0.24874 0.3522
0.17773 0.8052
0.34241 0.5623
0.31483 0.0809
0.00135 0.4287
0.27662 0.2051
0.01444 0.0001*
0.29810 0.3348
0.41104 0.0022*
0.35426 0.0103*
0.40831 0.0014*
0.45615 0.1392*
Pr >

OF Chi-Square

4 0.0149*

2 0.1934

Risk
Ratio

0.970
0.803
1.372
0.793
1.045
1.219
1.733
1.001
1.420
0.941
0.750
3.529
2.481
3.676
1.963

 

* Used in reduced model

116

Table 11b Length of Time Until Court Sanctioned Technical Violations:
Reduced Model

Summary of the Number of Event and Censored Values

Percent
Total Event Censored Censored
201 81 120 59.70
Testin lobalNullH othesis:BETA=0

Without With

Criterion Covariates Covariates Model Chi—Square

-2 LOG L 774.251 718.219 56.032 with 5 DF (p=0.0001)
Score 35.559 with 5 DP (p=0.0001)
Wald 32.196 with 5 BF (p=0.0001)

Analysis of Maximum Likelihood Estimates

 

Parameter Standard Pr > Risk
Variable DF Estimate Error Chi-Square Ratio
Treatment Weeks 1 -0. 060868 0. 01 302 0.0001* 0.941
SPCC 1 1.138381 0.36125 0.0016* 3.122
PES 1 0.926838 0.32841 0.0048* 2.527
SPCC/PES 1 1 .289301 0.36640 0.0004* 3.630
BC/PES 1 0.780416 0.44115 0.0769 2.182
Linear Hypotheses Testing

Wald Pr >
Label Chi-Square DF Chi-Square
COHORT CATEGORY 14.8349 4 0.0051*
* p < .01

**p<.05

117

offenders are often faced with severe consequences for technical violations
and are removed from community supervision for considerable lengths of
time. This presents limitations in measuring subsequent felony arrests
when survival analysis methods are used to measure recidivism for new
felony crimes within the follow up period. As previously discussed, survival
analysis assumes a well-defined end point and recurring events should not
be analyzed with this methodology. When offenders are rearrested and
receive additional court sanctions that remove them from community
supervision, they have by definition failed, even though they may return to
community supervision after 60 days or so. In a survival analysis, this
failure is treated as a terminal end point and the case is not included in the
accumulative proportion still surviving at the next time interval. Cases can
not be reentered in the analysis at a later time to measure a reoccurring
event (in this case a rearrest for a new felony) following reentry into the
community. It is therefore not possible to follow offenders who have failed
due to a technical violation past the technical violation failure date to
determine if a future felony crime was committed. In the following analysis
technical failures are censored events at the time of the technical failure
and not considered in the accumulative proportion still at risk for felony
failure past the technical failure date. Moreover, the proportion of censored

events to non-censored events is considerable. Consequently, the rate of

118

subsequent felony crime may be under or over estimated. And the degree
to which felony crime estimations are affected will be much greater in the
cohorts which have the largest proportion of offenders who fail in the
survival models as a consequence of technical violations. With this caveat,
any conclusion drawn from the interpretation of the following analysis must
be considered with caution.

Another Cox regression was run to examine the risk of failure for new
felony crime. Again, all independent variables were first tested prior to
running the reduced model (Table 12a). Table 12b provides the results of
the reduced Cox regression for new felony crimes (refer back to the
survival analysis plot, Fiqure 3, for visual illustration). It continues to be
documented that the single best predictor of outcomes is the length of time
in substance abuse treatment (p<.0025). Offenders who remain in
treatment are less likely to commit a subsequent felony crime. Also related
to felony crime failure is education level. Specifically, the higher the
education the less likely the offender is to recidivate. Offence category is
shown to have a significant effect on recidivism as well. Offenders
sentenced for drug crimes are 3.4 times more likely to commit new felony
crimes over time compared to offenders sentenced for violent crimes
(p<.007). In addition, the offenders’ drug of choice affect failure in this

model. Offenders who report their drug of choice to be cocaine/crack or

119

Table 12a

Length of Time Until New Felony Crimes: Full Model

 

Summag of the Number of Event and Censored Values

Percent
Total Event Censored Censored
201 41 160 79.60

Testin lobalNullH othesis: BETA=0
Without With

Criterion Covariates Covariates Model Chi-Square
-2 LOG L 395.368 353.615

Score

Wald

41.753 with 15 DF (p=0.0002)
35.225 with 15 DF (p=0.0023)
29.013 with 15 DF (p=0.0160)

Analysis of Maximum Likelihood Estimates

Parameter Standard Pr >
Variable DF Estimate Error Chi-Square
Age 1 -0.009977 0.02407 0.6785
Race 1 -0.244977 0.37301 0.51 13
Education Level 1 -0.792231 0.38498 003%“
Employment 1 0.306097 0.34730 0.3781
Prior Felonies 1 0.236929 0.23619 0.3158
Offence/Property 1 0.358149 0.53378 0.5022*
Offence/Drug 1 1 .306179 0.50173 0.0092*
Jail Days 1 0.000012410 0.00194 0.9949
Drug of Choice 1 0.645601 0.39252 0.1000*
Treatment Weeks 1 -0.055025 0.01918 0.0041*
Probation Violation 1 0.1 12387 0.43144 0.7945
SPCC 1 0.405433 0.50398 0.421 1*
PES 1 -0.350006 0.48081 0.4666*
SPCC/PES 1 -0.565544 0.69627 0.4166*
BC/PES 1 0.139267 0.54381 0.7979*
Linear Hypotheses Tasting

Wald Pr >

Label Chi-Square DF Chi-Square
COHORT CATEGORY 3.3628 4 0.4990*
OFFENCE CATEGORY 8.3948 2 0.0150*

Risk
Ratio

0.990
0.783
0.453
1.358
1.267
1.431
3.692
1.000
1.907
0.946
1.119
1.500
0.705
0.568
1.149

 

* Used in reduced model

120

Table 12b Length of Time Until New Felony Crimes: Reduced Model

 

Summary of the Number pf Event and Cepsored yalues

Percent
Total Event Censored Censored
201 41 160 79.60

Tasting Clpbal Null Hypothesis: 8ETA=0

Without With
Criterion Covariates Covariates Model Chi-Square

 

-2 LOG L 395.368 355.730 39.638 with 9 DF (p=0.0001)

Score . . 32.117 with 9 DP (p=0.0002)
Wald . . 27.588 with 9 DP (p=0.0011)
An I sis of 'mum ikel'hood E timates

Parameter Standard Pr > Risk
Variable DF Estimate Error Chi-Square Ratio

-0.761 1 15 0.34903 0.0292“r 0.467
0.276610 0.47857 0.5633 1.319
Offence/Drug 1 .230864 0.45943 0.0074* 3.424
Drug of Choice 0.780170 0.36246 0.0314” 2.182

Education Level 1
1
1
1
Treatment Weeks 1 -0.053420 0.01765 0.0025* 0.948
1
1
1
1

Offence/Property

SPCC 0.430847 0.46786 0.3571 1 .539
PES -0.295762 0.44967 0.5107 0.744
SPCC/PES -0.634060 0.66455 0.3400 0.530
BC/PES 0.219816 0.52437 0.6751 1.246

in H theses Testin

Wald Pr >
Label Chi-Square DF Chi-Square
COHORT CATEGORY 3.7206 4 0.4451
OFFENCE CATEGORY 8.9649 2 0.0113*

 

* p<.01
**p<.05

121

polydrug are 2.1 times more likely to recidivate compared to those whose
drug of choice is alcohol or THC. Finally, offenders sentenced to
residential probation/tether (SPCC/PES) and tether (PES) appeared to do
just slightly better, while offenders sentenced to residential probation
(SPCC) and boot camp/tether (BC/PES) appear to be somewhat more
likely to recidivate when compared to offenders sentenced to standard
probation. These differences are not statistically significant. Again, the
proportion of technical violations (censored data) in this model and the
small number of observation within each cohort presents limitations in
drawing conclusions relative to the relationship between probation models
and subsequent felony crime. Thus, hypothesis 3b which states; “the risk
of failing over time for new felony crimes over time is less among offenders

sentenced to residential probation” can not be supported with these data.

Substance Abuse Treatment Access and Retention:

 

Even with a small sample, what continues to be very significant in all
models is the impact that substance abuse treatment has on outcomes.
Research consistently shows compulsory treatment to be an effective
means in achieving positive outcomes. How a substance abuser is
exposed to treatment seem irrelevant. What is important is that drug

involved offenders are brought into an environment where treatment can

122

occur (Maddux, 1988; Hubbard, et al., 1988; McGlothlin, et al.,1977a;
1977b). Given the importance of this link, ensuring compliance with court
ordered treatment is therefore the first step in achieving positive outcomes.
Unfortunately, a significant proportion of offenders court ordered to
treatment are discharged from probation before having complied with court
ordered mandates. Research has also found that offenders sentenced to
more intensive probation programs are more likely to participated in
treatment compared to offenders sentenced to standard probation
(Petersilia and Turner, 1993a). Given the structure of the residential
probation program it was believed that the residential probation program
evaluated in this study was best able to ensure compliance with court
mandated treatment compared to other probation models.

To test this assumption, data were analyzed using a logistic
regression model. Again, this method was selected because it quantifies
the association between risk factors and an event after adjusting for other
variables as relative risk ratio, but requires the outcome variable to be a
binary or dichotomous measure as opposed to a time sensitive measure.
Here, the evaluation designed to determine whether offenders sentenced to
residential probation are more likely to access substance abuse treatment.
Thus, the outcome measure is quantified as dichotomous. Offenders either

enter treatment or they did not. The same process used in constructing the

123

Cox regression models was employed for the logistic regression. All
independent variables were entered into an initial analysis (Table 13a)
followed by a reduced model which includes only those variables noted as
significant or approaching significance in the initial analysis. Table 13b
provides the results of the reduced logistic regression model. Clearly
offenders sentenced to both residential probation program (SPCC and
SPCC/PES) do better with respect to this outcomes measure. Offenders
sentences to SPCC are 4.9 times more likely to enter treatment and
offenders sentenced to SPCC/PES are 3.8 times more likely to enter
substance abuse treatment compared to offenders sentenced to standard
probation. The strength of the relationship is statistically significant

(p = .002 and p = .014). Offenders sentenced to tether (PES) and boot
camp/tether (BC/PES) are some what less likely to enter treatment,
although the association is not statistically significant. Additionally, Table
13b indicates that the length of time offenders spend in jail as a
consequence of the critical offence, significantly affects whether offenders
enter treatment or not (p =.029). The relationship is inverse; the longer
offenders are incarceration the less likely they are to enter treatment
following release. The overall offence category is shown to have a
significant effect on treatment access as well (p =.02). When drug and

property crimes are compared to the hold out category (violent crimes) it is

124

Table 1 3a

Offenders Receiving Substance Abuse Treatment: Full Model

 

Response Profile

Ordered
Value Substance Abuse Treatment Count
Yes 1 118
No 0 83
Testin lobal Null H othesis B TA=0
Intercept
Intercept and
Criterion Only Covariates Chi-Square for Covariates
-2 LOG L 271.451 231.103 40.348 with 14 DF (p=0.0002)
Score 36.899 with 14 DF (p=0.0008)
Analysia of Maximpm Likelihood Estimatas
Parameter Standard Pr > Odds
Variable DF Estimate Error Chi-Square Ratio
lNTERCPT 1 -0.2120 0.7184 0.7679 .
Age 1 0.00477 0.0206 0.8170 1.005
Race 1 0.4883 0.3659 0.1820 1 .630
Education Level 1 0.2624 0.3585 0.4642 1.300
Employment 1 0.1954 0.3477 0.5742 1.216
Prior Felonies 1 -0.0166 0.2101 0.9371 0.984
Offence/Property 1 0.3405 0.4390 0.4379* 1 .406
Offence/Drug 1 -0.8335 0.4555 0.0673* 0.435
Jail Days 1 -0.00448 0.00197 0.0230* 0.996
Drug of Choice 1 0.0544 0.3497 0.8764 1.056
Probation Violation 1 0.2867 0.4322 0.5071 1.332
SPCC 1 1.7094 0.5339 0.0014* 5.526
PES 1 -0.0862 0.4395 0.8444* 0.917
SPCC/PES 1 1.3673 0.5494 0.0128* 3.925
BC/PES 1 -0.4550 0.6464 0.4814* 0.634
Linaar Hypotheses Testing
Wald Pr >
Label Chi-Square DF Chi-Square
COHORT CATEGORY 18.6121 4 0.0009*
OFFENCE CATEGORY 7.7197 2 0.0211*

 

* Used in reduced model

125

Table 13b Offenders Receiving Substance Abuse Treatment:
Reduced Model

 

Response Profile

Ordered

Value: Substance Abuse Treatment Count
Yes 1 118
No 0 83

Testin IobalNullH othesis BETA=0

Intercept
Intercept and
Criterion Only Covariates Chi-Square for Covariates
-2 LOG L 271.451 235.547 35.904 with 7 DF (p=0.0001)
Score . . 33.264 with 7 DF (p=0.0001)

Analysis of Maximum Likelihood Estimates

 

Parameter Standard Pr > Odds
Variable DF Estimate Error Chi-Square Ratio
Intercept 1 0.3754 0.4334 0.3864
Offence/Property 1 0.3386 0.4181 0.4180 1 .403
Offence/Drug 1 -0. 7081 0.41 09 0.0848 0.493
Jail Days 1 -0.00407 0.00187 0.0297“ 0.996
SPCC 1 1.5805 0.5198 0.0024* 4.857
PES 1 -0.0210 0.4080 0.9590 0.979
SPCC/PES 1 1.3257 0.5398 0.0141 * 3.765
BC/PES 1 -0.7151 0.6290 0.2555 0.489
Linear Hypotheses Testing

Wald Pr >

Label Chi-Square DF Chi-Square
COHORT CATEGORY 19.0858 4 0.0008
OFFENCE CATEGORY 7.8043 2 0.0202*
* p < .01
** p < .05

126

noted that offenders sentenced for property crimes are somewhat less
likely to enter treatment and offenders sentenced for drug crimes are 1.5
more likely to enter treatment. This multivariant analysis supports the
bivariant analysis discussed previously and hypothesis 1 which states,
“probationers participating in residential probation are more likely to
participate in substance abuse treatment.”

Getting offenders into treatment is only the first step. Keeping them
there for periods of time long enough to bring about change is another
issue. Several studies well document that the length of time offenders
remain in treatment is the most consistent and best predictor of treatment
outcomes. This is true even among those who drop out of treatment
(Simpson and Friend, 1988; DeLeon, 1988, 1994; Hubbard et al., 1988;
Barr and Antes, 1981). It was the intent in this evaluation to measure the
effect of residential probation on treatment retention (hypothesized 2a;
“Drug-involved offenders sentenced to residential probation will remain in
substance abuse treatment longer”). In order to test the effect of residential
probation on treatment retention the actual length of time all offenders were
expected to remain in treatment had to be known. This information was
only really known for those offenders sentenced to the residential
probation. Some assumptions could have been made with respect to

offenders in the other cohorts, but the probability of error in making any

127

assumption was too great. In addition, treatment expectations would have
to be consistent among all offenders in order to establish true differences in
treatment retention rates between cohorts. This was not the case. The
length of time offenders actually spent in treatment appeared, to some
degree, to be a function of the treatment provider and where the services
were delivered. For example, the treatment provider delivering services at
the residential probation program provided two phases of a three phase
out-patient treatment program. This is a function of the average length of
time offenders spend in the residential probation program. Upon
successful completion of the residential probation program and successful
completion of the two phase substance abuse treatment program,
offenders are successfully discharged from both residential probation and
treatment. In contrast, offenders in other probation models receiving
treatment services from the same provider at the provider’s out-patient
location may be expected to complete all three phases of the out-patient
program. In addition, a meaningful proportion of offenders received
outpatient services from other providers in the county where treatment
expectations range from one month to unknown. Therefore, treatment
retention is not evaluated, but instead differences in the actual length of
time offenders remained in treatment are examined. This does not suggest

that the information derived from this analysis is without value. The actual

128

length of time offenders remain in treatment is consistently shown to effect
other outcome measures.

To examine this issue, the actual experience of all offenders who
entered treatment, both those who dropped out of treatment and those who
completed treatment is examined and comparisons between cohorts are
made. First, a univariant survival analysis was conducted and descriptive
statistics produced using SAS Life time test function (SAS Institute Inc.,
1997 pp. 413-431) to illustrate differences the length of time offenders
remained in treatment between cohorts. Survival analysis was used
because the data are time sensitive however, there were no censored
observations in this analysis. All offenders in the study sample who
entered treatment terminated from treatment by the end of the observation
period. Moreover, the concept of treatment failure has an ambiguous
relationship to length of time in treatment because treatment expectations
among offenders varied considerably. Using treatment failure as a
censored observation has little meaning and is therefore not appropriate.
Thus, the univariant survival analysis was construct using all offenders who
entered treatment and plots the length of time offenders in each cohort
were in treatment from date of admission until the date of termination,
regardless of the reason for discharge. This analysis illustrates only the

actual length of time offenders stayed in treatment. No conclusions

129

regarding treatment success or failure can be drawn (refer back to Figure
4). As shown, the average length of time offenders remained in treatment
ranged from a mean of 14.6 weeks to 28.5 weeks. ANOVA statistics were
then used to test the observed differences in the mean length of time in
treatment between cohorts. The results of the ANOVA test indicate that
observed differences are not statistically significant (p = 0.33, see Table
14). Hence, Bonferroni or Scheffe’s test were not indicated and therefore
not performed. Although this analysis can not support hypothesis 2a, the
data does illustrate that once offenders get into treatment, the length of

time they remain in treatment is fairly consistent across the study sample.

Table 14 Weeks in Substance Abuse Treatment: ANOVA Results

 

 

Std.Prob SPCC PES SPCC/PES BCIPES
Mean 28.5 18.5 28.8 22.9 14.6
Median 21.0 12.0 23.0 16.5 8.0
Minimum 1.0 3.0 2.0 2.0 1.0
Maximum 105.0 109.0 90.0 109.0 41.0
SE 4.5 4.0 4.7 4.9 7.1

ANOVA P-VALUE: 0.33

 

Std.Prob = standard probation; SPCC = residential probation; PES = tether;
SPCCIPES = residential probation with tether; BCIPES = boot camp with tether

Alcohol and Drug Relapse:

The analysis continues by examining drug and alcohol relapse

patterns and testing the relationship of the independent variables to the

130

length of time offenders remained substance free. Offenders are typically
screened once a month for alcohol, marijuana (THC) and cocaine. If
indicated, offenders are screened for other types of drugs. Because it was
determined that drug relapse for cocaine and other drugs are of greater
concern among officials in the study jurisdiction and therefore, court
sanctions are more likely to occur for subsequent cocaine/ other drug use
compared to relapse for alcohol and THC, it was important to analyzes
relapse for alcohol and THC use separately from relapse for cocaine and
other drugs. Consequently, THC use was not considered in the drug
category. Instead, alcohol and THC use were collapsed into one category.
The “drugs other than alcohol/THC” category includes cocaine/crack
cocaine and poly-drug use. These two categories, “alcohol/THC and
“other drugs”, are examined separately.

Like recidivism data, these data are time sensitive and include
censored observations. Survival analysis methods are the most appropriate
method for analyzing these data and, to examine the length of time to
failure for each category requires censoring the event that is not of interest.
However, before constructing the survival analysis the data were checked
to determine the likelihood that failure due to other drugs would precede
failure due to alcohol and THC. This was considered important because it

would be misleading with respect to substance abuse relapse to examine

131

alcohol and THC relapse separately (censoring other drug use) if a
significant proportion of offenders were failing due to using other drugs first.
This event occurred in only 3 cases. Therefore, it was determined that
examining alcohol/THC use separately and censoring out failure due to
other drug use, would be appropriate and the analysis would yield an
accurate assessment of substance abuse relapse with respect to alcohol
and TCH use. In each analysis, both univariant and multivariant survival
methods are used. First, alcohol and THC relapse is examined.

Figure 5 and Table 15 illustrates the univariant survival analysis and
descriptive statistics for alcohol and THC relapse. As shown, although
there does not appear to be marked differences in mean survival times
between cohorts, there are notable differences in the proportion of
offenders who remained alcohol and THC free in the first year.
Approximately 78% of offenders in both residential probation programs
(SPCC and SPCC/PES) remained alcohol and THC free for approximately
35 week compared to only about 64% of offenders supervised on tether
(PES) and 56% supervised in standard probation. More than half of the
BC/PES cohort had relapsed during this time period. The appreciable
difference dissipates at about 52 weeks. At around one year relapse
stabilizes among all cohorts. This pattern is very consistent with the

substance abuse relapse literature. Substance abusers are most

132

of

r

c:

_

of

_

mma—Gm o

O: o:

_

 

_

52.22. 2.: 95 .282 2.235 9.33

mmfioomm o mma 4
me E 9.33

E: om cm 3 o

_ _ _ _

 

’>

comm o

g cm. 3 on

r _ _ _

O

 

 

 

 

 

 

 

—4I

 

m 2:9“.
5:29.“. .30. .4

E E Q

_ b _

g
-2
.2
-2

L:

 

 

 

 

 

 

 

.l.
e
_..
r .2

my too

-2:

 

 

Bumuuns %

133

Table 15 Weeks Surviving Alcohol and THC Relapse

 

 

A = Std. Prob B = SPCC C = PES D = SPCC/PES E = BC/PES
Mean 74.0 75.6 80.9 75.1 44.8
Median 54.1 59.1 * 58.9 28.0
SE 8.9 10.7 9.5 10.6 12.8

More than 50% still surviving at the end of the follow-up period

 

vulnerable to relapse during post acute or late physical withdrawal which
reaches its peak at 3 to 6 months and becomes less intense at about one
year (Fitzgerald, 1993).

To examine the differences, Cox regression methods are used. The
same process used in constructing all other Cox regression models was
employed. All independent variables were entered into an initial analysis
(Table 16a) followed by a reduced model which includes only those
variables noted as significant or approaching significance in the initial
model. Table 16b provides the results of the reduced model for alcohol
and THC relapse. Again, length of time in treatment continues to be
significant in predicting outcomes (p=.0083). The longer offenders remain
in treatment the less likely they are to relapse. Age however, is noted as
the most significant predictor of failure in this model (p=.0003).
Specifically, older offenders are less likely to fail due to alcohol or THC use
than younger offenders. In addition, there appears to be a relationship

between the length of time offenders are incarcerated as a result of the

134

 

Table 16a Length of Time Until Alcohol or THC Relapse: Full Model

 

mm f the Number f vent and ns re V l es

Percent

Total Event Censored Censored

170 76 94 55.29

Testin lobal Null H othesis: TA=0

Without With

Criterion Covariates Covariates Model Chi-Square

-2 LOG L 699.074 658.357 40.717 with 15 DF (p=0.0004)
Score 34.184 with 15 DF (p=0.0032)
Wald 33.945 with 15 DF (p=0.0035)
Analysis of Maximum Likelihood Estimates

Parameter Standard Pr > Risk
Variable DF Estimate Error Chi-Square Ratio
Age 1 -0.050278 0.01999 0.0119* 0.951
Race 1 -0.292413 0.26048 0.2616 0.746
Education Level 1 0.124212 0.27764 0.6546 1.132
Employment 1 -0.390530 0.26325 0.1379 0.677
Prior Felonies 1 -0.01 1776 0.18648 0.9496 0.988
Offence/Property 1 0.240151 0.34827 0.4905 1.271
Offence/Drug 1 0.454272 0.32700 0.1648 1.575
Jail Days 1 -0.002859 0.00165 0.0832* 0.997
Drug of Choice 1 -0.287550 0.27543 0.2965 0.750
Treatment Weeks 1 -0.017206 0.00706 0.0148* 0.983
Probation Violation 1 -0.036147 0.32098 0.9103 0.964
SPCC 1 -0.556241 0.39006 0.1539* 0.573
PES 1 -0.064555 0.331 10 0.8454* 0.937
SPCC/PES 1 -0.752353 0.42856 0.0792* 0.471
BC/PES 1 -0.313161 0.48673 0.5200* 0.731
Linsar Hypotheses Testing
Wald Pr >

Label Chi-Square DF Chi-Square
COHORT CATEGORY 4.5917 4 0.3318*

OFFENCE CATEGORY 1.9352 2 0.3800

 

* Used in reduced model

135

Table 16b Length of Time Until Alcohol or THC Relapse: Reduced Model

 

Summam of the Number of Event and Censorad ValtJes

Total Event
1 70 76

Percent

Censored Censored
94

55.29

Testin IobalNulIH othesis: BETA=0

Without
Criterion Covariates Covariates Model Chi-Square

-2 LOG L 699.074

Score
Wald

With

664.870 34.204 with 7 DP (p=0.0001)
. 27.773 with 7 DF (p=0.0002)
27.291 with 7 DP (p=0.0003)

Analysis of Maximum Likelihood Estimates

Variable

Age

Jail Days
Treatment Weeks
SPCC

PES

SPCC/PES
BC/PES

DF

AA—L—k—X—l—b

Parameter Standard Pr > Risk
Estimate Error Chi-Square Ratio

-0.061 170 0.01697 0.0003* 0.941
-0.002794 0.00157 0.0747” 0.997
-0.017622 0.00667 00083" 0.983

-0.554666 0.35422 0.1 174 0.574
-0.175018 0.31367 0.5769 0.839
-0.693848 0.39414 0.0783” 0.500
-0.375819 0.45513 0.4089 0.687

Linear Hypotheses Testing

 

Wald Pr >
Label Chi-Square DF Chi-Square
COHORT CAREGORY 4.4353 4 0.3503
*p < .01
**p < .05
*** p < .10

136

critical of offence, and failure. Although not significant at the .05 level the
relationship is approaching significance (p = .07) and seem strong enough
to indicate that a larger sample would yield greater significance. This
relationship is present in only one other analysis (treatment access) and is
therefore particulary interesting. Biological and physiological aspects of
addiction are often over looked when attempts are made to explain
behavior but may offer a plausible explanation in this case. This possible
explanation is discussed in detail in Chapter VI.

Finally, offenders sentenced to residential probation (SPCC) and
residential probation with tether (SPCC/PES) appear to do slightly better
over time in maintaining abstinence from alcohol and TCH, compared to
offenders sentenced to standard probation. While these differences are
not statistically significant (p=.11 and p=.07 respectively) the observed
change is in the direction hypothesized.

The analysis for drug relapse replicates the above process. A
univariant survival analysis was conducted to illustrate differences in the
length of time offenders remained drug free between cohorts. However, the
survival model was constructed slightly differently. This analysis did not
censor the event that is not of interest; alcohol and THC failure. The
rationale for this is straight forward. Many offenders use both marijuana

and cocaine or other drugs. A review of the data showed that a meaningful

137

proportion of offenders who relapsed due to cocaine had previously tested
positive for THC. If the analysis was constructed to censor alcohol/THC
failure, then analysis would clearly underestimate drug relapse.
Specifically, offenders who tested positive for THC prior to falling for
cocaine use would be censored from the analysis and the failure due to
cocaine would not be included. Therefore, all offenders are included in the
analysis until they fail due to a positive drug screen other than alcohol or
THC29 or until censored. As in prior analyses, censored events include,
death, lost to follow-up, transferred to another jurisdiction or removed from
community supervision (court sanctioned failure).

Although not as pronounced, the unitivariant survival analysis for
drug relapse reflects a similar pattern to that of the survival analysis for
alcohol/THC relapse (Figure 6 and Table 17). The proportion of offenders
remaining drug free is slightly greater among offenders sentenced to both
residential probation programs (SPCC and SPCC/PES) during the first few
months. Over 90% of offenders in both residential probation programs
remained drug free for a little more than 5 months compared to
approximately 83% of offenders sentenced to standard probation. Less

than 80% of offenders supervised on tether (PES) and approximately 65%

 

29

Not all probationers’ files contained drug screen information. These cases were
excluded from the analysis.

138

2:. ..o .282 :2: .25 3%.»: 9.5 9.335 9.33 o 2.5m

mmfioomm o mm.— 4 001m 0 5:32.— .Em 3..
:2: E 9.33
S: cm 3 3 ON 0

L _ _ _ _ _

-E
L:
L:
A:

Lie,

 

 

BulAuuns %

L;

 

 

 

 

 

 

 

 

L:

 

 

L;

A;

 

Lz:

 

139

Table 17 Weeks Surviving Drug Relapse Other Than Alcohol or THC

 

 

A = Std. Prob B = SPCC C = PES D = SPCC/PES E = BC/PES
Mean 102.8 91.3 98.8 94.7 62.2
Median * * * 120.1 80.9
SE 8.8 11.5 9.1 9.9 13.8
* More than 50% still surviving at the end of the follow-up period

 

of probationers in the boot camp/tether cohort remained drug free up to this
time period. These differences diminish rapidly but again, relapse stabilizes
at about one year.

Another Cox regression was run to examine the risk of failure for
drug use. Again, all independent variables were first tested prior to running
the reduced model (Table 18a). Table 18b provides the results of the
reduced Cox regression for drug relapse. The proportion of censored data
and the actual number of failure in this model undoubtedly affects this
analysis. Despite this limitation, length of time in treatment still is shown
to be significant in predicting outcomes (p=.03). Unlike failure due to
alcohol and THC use, the offenders’ drug of choice is also significant in
predicting failure due to drugs other than alcohol and THC. With respect to
the different probation models, no appreciable difference are noted. The
risk ratios do however indicate that observed changes are in the direction
hypothesized. Given this consistent pattern, it is expected that a larger

sample would yield true differences between cohorts.

140

Table 18a Length of Time Until Drug Relapse: Full Model

 

Summam of the Number of Event and Censored Values

Percent
Total Event Censored Censored
165 51 1 14 69.09

Testin lobalNullH othesis: BETA=0
Without With

Criterion Covariates Covariates Model Chi-Square

-2 LOG L 465.542 419.416 46.126 with 15 DF (p=0.0001)
Score . . 47.822 with 15 DF (p=0.0001)

Wald . . 40.035 with 15 DF (p=0.0004)
Analysis of Maximum Likelihood Estimates

Parameter Standard Pr > Risk
Variable DF Estimate Error Chi-Square Ratio
Age 1 0.023278 0.01955 0.2338 1.024
Race 1 -0.519975 0.3501 1 0.1375 0.595
Education Level 1 0.150241 0.35526 0.6724 1.162
Employment 1 0.286177 0.31056 0.3568 1.331
Prior Felonies 1 -0.038449 0.18936 0.8391 0.962
Offence/Property 1 -0.340204 0.38709 0.3795 0.712
Offence/Drug 1 0.510382 0.41319 0.2167 1.666
Jail Days 1 0.000033039 0.00171 0.9846 1.000
Drug of Choice 1 1.868225 0.41633 0.0001* 6.477
Treatment Weeks 1 -0.013608 0.00789 0.0845* 0.986
Probation Violation 1 -0.176929 0.42906 0.6801 0.838
SPCC 1 -0.164587 0.52460 0.7537* 0.848
PES 1 -0.046484 0.39532 0.9064* 0.955
SPCC/PES 1 -0. 192254 0.48786 0.6935* 0.825
BC/PES 1 0.471328 0.61205 0.4413* 1.602
Linaas yppthasss Tasting
Wald Pr >

Label Chi-Square DF Chi-Square
COHORT CATEGORY 1.0991 4 08944"
OFFENCE CATEGORY 4.1416 2 0.1261

 

* Used in reduced model

141

Table 18b Length of Time Until Drug Relapse: Reduced Model

 

Summam of the Number of Event and Censored Values

 

Percent
Total Event Censored Censored
165 51 1 14 69.09

Testin IobalNullH othesis: BETA=0

Without With

Criterion Covariates Covariates Model Chi-Square

-2 LOG L 465.542 428.750
Score
Wald

35.792 with 5 DF (p=0.0001)
35.357 with 5 DF (p=0.0001)
29.042 with 5 DF (p=0.0001)

Analysis of Maximum Likelihood Estimates

Parameter Standard Pr > Risk
Variable DF Estimate Error Chi-Square Ratio
Treatment Weeks 1 -0.01 5893 0.00740 0.031 7** 0.984
Drug of Choice 1 1.894012 0.37545 0.0001* 6.646
SPCC 1 -0.365963 0.43562 0.4009 0.694
PES 1 -0. 125840 0.36190 0.7281 0.882
SPCC/PES 1 -0.265091 0.43566 0.6047 0.748
BC/PES 1 0.299792 0.56795 0.5976 1 .350
Linear Hypotheses Testing

Wald Pr >
Label Chi-Square DF Chi-Square
COHORT CATEGORY 1.3744 4 0.8486

 

* p<.01
**p<.05

142

Summary:

To recapitulate, some of the findings discussed in this section were
consistent with the hypotheses outlined while other were not or could not
be adequately tested. As hypothesized, offenders sentenced to residential
probation are more likely to comply with court ordered substance abuse
treatment compared to offenders sentenced to less intensive probation
programs. This finding supports previous research which suggest that
offenders sentenced to more intensive probation programs are more likely
to participate in compulsory treatment. Interestingly, this analysis also
shows an inverse relationship between the length of time offenders spend
in jail prior to being released to community supervision and participation
with court ordered substance abuse treatment. The longer offenders are
incarcerated the less likely they are to enter treatment.

It was also hypothesized that offenders sentenced to residential
probation would remain in treatment longer. This was not the case.
Offenders sentenced to residential probation remained in treatment
approximately 6 - 10 weeks less compared to offenders sentenced to less
intensive probation programs (excluding BC/PES). However, this appears
to be a function of the length of time offenders are expected to remain in
treatment as opposed to a failure in retaining offenders in treatment.

Although not statistically significant, comparisons of the relative risk

143

ratios indicate that the likelihood of remaining alcohol and drug free is
slightly greater for offenders sentenced to residential probation compared
to offenders sentenced to less intensive probation programs. This is also
consistent with the hypotheses. The observed patterns show that a greater
proportion of offenders who were sentenced to residential probation remain
alcohol and drug free up to approximately one year, at which time
appreciable differences dissipate. Moreover, the analyses show that the
length of time offenders remain in treatment is a significant predictor of
alcohol and drug relapse which underscores the importance of ensuring
compliance with court mandated treatment.

Based on previous research it was anticipated that court sanctioned
technical violations would be greater among offenders sentenced to
residential probation. The data in this study support this assumption and
associated hypothesis. The likelihood of failure due to technical violations
is greatest among offenders sentenced to the most intensive probation
programs and least among offenders sentenced to the least intensive
probation programs. Again, the length of time offenders remain in
treatment is shown as the single best predictor of failure. This analysis
continues to underscore the importance of ensuring that offenders comply
with court mandated treatment and further, that offenders remain in

treatment long enough for treatment to bring about change.

144

This program evaluation was also designed to examine the effects of
residential probation on subsequent criminal behavior or recidivism.
However the proportion of court sanctioned technical violations and the
small number of subsequent felony crime convictions limited the ability to
draw meaningful conclusion from this analysis and also affected the ability
to test the following hypothesis which states: “Of the probationers
rearrested for new crimes, the severity of the new crime will be less for
offenders sentenced to residential probation.” Even though the sample
size was limited, the analysis conducted to measure the effects of
residential probation on recidivism still show time in treatment to be a

significant predictor of successful out comes.

145

VI. CONCLUSION AND DISCUSSION

Overview:

The greatest rise in incarceration in the country’s history has taken
place in the past three decades, contributing to significant prison
overcrowding. In response to this pressure, the use of probation has
expanded considerably causing an increase in the probation population of
almost 300% over the past 10 years. Not only is the probation population
growing in size, the type of offender on probation has changed as well.
More of the current population consists of offenders convicted of felonies
and a greater proportion of offenders are drug dependent. Because
standard probation was never intended to manage these types of
offenders, over the past decade many models of intermediate sanctions or
intensive supervision probation programs were developed to cope with
probation’s dilemma. Critics of prison diversion are many, arguing that
diverted offenders continue to commit crimes and rarely reform.
Meanwhile supporters reinforce the many advantages of probation and
argue the difficulty lies in its execution, not its theory.

As the debate continues, many studies have been conducted in
efforts to determine the effectiveness of various ISP programs models and

identify program characteristics which best predict improved client

146

outcomes. Research has shown that monitoring alone is not effective in
producing positive outcome, but rather successful programs are those that
combine substance abuse treatment with surveillance. What remains
unknown are the specific levels of both treatment and monitoring required
to maximize outcomes and for particular types of offenders. The program
evaluation described in this dissertation was an attempt to respond to this
need and examined the effectiveness of a different and unique structure of
ISP (residential probation) in combination with a very specific level of
substance abuse treatment (out patient) for drug involved felony offenders.
The literature was first reviewed to determine key risks factors or
predictors of recidivism. These risk factors determined the independent
variable and included, age, race, education level, employment, prior
felonies, offence category, drug of choice and substance abuse treatment.
In addition, the actual time offenders spent in jail as a consequence of the
incidence offence and probation violation status were also included as
independent variables. Essentially, four outcomes measures were
considered: treatment access, length of time in treatment, subsequent drug
use or relapse, and recidivism. Both univariant and mulitvariant methods
were used to test the hypotheses. However, certain limitations were
encountered that restricted the analyses and the ability to fully test one of

the hypotheses.

147

Limitations:

The most significant limitation to this dissertation is the size of the
study population. To avoid sampling error, the study was design to utilize a
population-based cohort which included the total population of offenders
meeting the admission criteria for residential probation. Sample size
calculations were performed to ensured the study population was
adequate. Results indicated that the identified study cohort was 10%
larger than was needed. However, over 25% of the study cohort was lost
in the data collection process due to missing files or missing data.
Moreover, the high rate of subsequent technical violations was not
anticipated and further reduced the sample in some of the analyses. This
was not problematic in the bivariant analyses but presented limitations
when conducting some of the multivariant analyses. As a result, some of
the analyses were not able to provide certain information sought by the
officials and policy makers supporting the evaluation. The other
unexpected nuance was the significant proportion of technical violators
sentenced to residential probation compared to the other probation
programs. If the residential probation program is truly servicing more of
this population then a more accurate assessment of the program may be to
compare outcomes of technical violators across probation models. Finally,

the study was designed to evaluate the effectiveness of a specific program,

148

servicing a specific population and provide information to program
managers for the purpose of developing policy. Therefore the study is
limited in its generalizibility.

However, despite these limitations the results of the study provide
meaningful information which have implications for both the local criminal
justice and treatment agencies and for future research. A discussion of the
findings, recommendations and direction for future research are presented

in this section.

Discussion of Findings and Policy Recommendations:

Treatment Access

The established link between substance abuse treatment and
treatment outcomes among criminal justice clients is well document.
Consistently, length of time in treatment is found to be the single best
predictor of positive outcomes. Unfortunately, a sizable number of
offenders are discharged from probation before having complied with court
mandated treatment, indicating a gap between the criminal justice and
treatment systems. Thus, despite court mandates, getting offenders into
treatment remains an issue whether this is a function of treatment capacity
or offender noncompliance. This study was designed to determine the

effectiveness of residential probation in facilitating treatment access and

149

 

ensuring compliance with court mandated treatment. On this measure
residential probation is by far the most effective compared to other
probation programs. Offenders sentenced to residential probation were
almost 4 to 5 times more likely to enter substance treatment compared to
offenders sentenced to standard probation. This is consistent with previous
research which show that offenders who participate in more intensive
probation programs are more likely to participate in substance abuse
treatment and counseling. Moreover, this finding appears congruent with
deterrence theory. It is assumed that the “perceived threat” of punishment
and the “certainty of punishment” for failing to comply with court mandated
treatment are far greater for offenders sentenced to residential probation.
Therefore, greater treatment compliance rates would be expected.30
Whether these treatment compliance rates are a function of perceived
threat, the certainty of punishment or a function of on site treatment, the
residential probation program’s strategies seem to be effective in ensuring
that offenders comply with court mandated treatment. The importance of
this can not be overstated given the relationship between substance abuse
treatment and outcomes.

When analyzing “treatment access,” the length of time offenders

 

30

To provide an adequate test of deterrence theory or test this assumption, data regarding
the types of threats, the offenders’ perceptions of the threats and the actual consequences
(certainty of punishment) are needed. Only outcome data (treatment admission) was available.

150

spend in jail for the critical offences was shown to be a significant predictor
of this outcome measure. The longer offenders spent in jail the less likely
they were to enter treatment after release. Only speculation can be offered
to explain this relationship. Previous research in treatment retention
(Wexler, et al., 1990) suggests that greater exposure to treatment produces
a positive effect up to a point of satiation. After this point, positive
outcomes declined with more treatment, suggesting a dosage model. It is
possible that incarceration may have a similar effect on treatment
motivation. Incarceration may serve as a treatment motivator up to a point
of satiation and after this point the longer the confinement the less
motivated the offender may become. As pointed out in the drug treatment
literature, motivating clients to enter treatment is often problematic. Denial
of how severe the problem really is for the substance abuser can
significantly affect the clients’ motivation or “readiness for treatment.”
Emotional pain or the pain of a significant loss, which can occur as a
consequence of addictive behavior, can help reduce the denial and
resistance to treatment (Simpson and Marsh, 1986). According to Martin
(1982) the opportune time to motivate the substance abuser for treatment
is at the point when he is experiencing emotional pain as a consequence of
addictive behavior. However, if the negative consequences of the addictive

behavior are not reinforced the pain and motivation for treatment will

151

subside over time. When the pain subsides, the motivation is gone and the
' denial of the problem returns. It would then follow that the longer the
confinement, the farther removed the offender becomes from the painful
event or the treatment motivator. Given the importance of treatment, this
relationship is worth further investigation to determine if optimal levels of
incarceration exist for different offenders under various conditions.
Certainly the data suggests that there are.

Regardless of the reason, the relationship between length of
incarceration and the likelihood of treatment admissions is evident
throughout the study sample. To improve treatment admissions some form
of intervention either by the criminal justice system (i.e., probation officer)
or by the treatment provider is indicated for offenders incarcerated for long
periods of time. A well establish transition plan to link these offenders to
treatment directly from jail should be considered. For example, in a
jurisdiction known to this writer, when drug involved felony offenders are
released from the county jail they are transported by the county from jail
directly to out-patient treatment services. Offenders are not allowed to be
released to or transported by family members. On the same day offenders
are released from jail, they are admitted and required to participate in
treatment. This policy ensures the gap between the criminal justice and

treatment systems is closed. Another alternative may be to develop an

152

 

intervention or treatment program while the offender is incarcerated, that
uses the personal crisis which has significant personal implications for the
addict to guide and affirm change and keep the offender In a readiness
state for treatment. Other factors and processes from body of knowledge
in the medical sociology literature that facilitate help-seeking behavior and

or maintain the readiness for treatment state should be explored as well.

en th of Ti in Tre merit

This evaluation also was designed to examine the effects of
residential probation on treatment retention. However, this was not
possible. To explain, the distinction between length of time in treatment
and treatment retention is restated. Obviously, length of time in treatment
is simply the actual time offenders spent in treatment from the date of
admission to the date of discharge, irrespective of success or failure. In
comparison, treatment retention is not just how long the offender remains in
treatment, but also whether clients are retained in treatment until a
specified and point. Clients who drop out before the specified end point
are considered to have failed. To measure retention therefore requires that
the specified and point is known. Moreover, to compare the effectiveness
of one intervention with others in retaining offenders in treatment requires

the expected length of treatment to be consistent across interventions. lf

153

the expected length of treatment in one intervention is one month and in
another intervention it is four months, appropriate comparisons can not be
made. Within the study sample, the expected time in treatment was either
not known for many or varied considerably between offenders. To adjust
the study design to the realities of the field, the actual length of time
offenders remained in treatment were compared to determine if there were
meaningful differences between offenders sentenced to residential
probation and other probation programs.

The average length of time offenders sentenced to residential
probation stayed in treatment ranged from means of 18.5 to 22.9 weeks for
the different subgroups. The analysis show that these differences are not
statistically significant. However, it could be argued that the observed
difference may have practical significance. The treatment experience for
offenders sentenced to residential probation is approximately 6 to 10
weeks less compared to offenders sentenced to less intensive probation
models (excluding BC/PES). Given the relationship between length of time
in treatment and treatment outcomes, this variance could be important. It
has been documented that treatment durations of 3 months or more are
necessary to predict improved outcomes (Tims, et al., 1991).

The differences in time in treatment do not appear to be a retention

or treatment failure issue. Among offenders who actually entered

154

 

treatment, treatment success rates were compared. Bivariant analysis
shows that observed differences in the proportion of offenders successfully
completing treatment are not statistically significant between cohorts. This
suggests that the 6 to 10 weeks difference in the average length of time in
treatment observed among offenders sentenced to residential probation is
a function treatment expectations. Offenders sentenced to either
residential probation programs are expected to complete approximately 14
weeks of out-patient treatment while in residence. The 14 week program
consists of two phases of a three phase program which is offered by a local
treatment provider. Due to the average length of time offenders remain in
residential probation only two phases are provided at the program site.
After successfully completing residential probation and the two treatment
phases, offenders are discharged from residential probation and the sole
responsibility for supervision is transferred back to the adult probation
department. Many leave with a “recommendation” from the provider to
continue treatment and complete the third phase. Because of state
regulatory substance abuse treatment licencing rules, offenders are
discharged from treatment and considered successful completers at that
level. If offenders are to continue treatment, even with the same provider,
they must be readmitted at the out-patient location. In theory, offenders

have completed treatment and the offenders’ perception may be the same.

155

It appears that many offenders discharged from residential probation do not
continue on with treatment and there does not appear to be reason for the
offender to do so. In contrast, offenders sentenced to standard probation
or tether and receiving services from the same provider attend treatment at
the provider’s primary out-patient treatment facility. Offenders are admitted
with the expectation that they continue through the three phases and can
do so without any interruption in services between phases. This would
account for the 6 to 10 week variance in the average length of time in
treatment.

Clearly the residential probation program is far more effective in
engaging offenders in treatment process. Moreover, the proportion of
offenders who successfully complete treatment is equivalent compared to
other cohorts. However, the average length of time offenders remain in
treatment is notably less which appears to be a system function. A gap
between the criminal justice treatment system is created when offenders
are successfully discharged from residential probation and treatment then
continued on standard probation with a recommendation to continue
treatment. To close this gap, a strategy to motivate offenders to continue
treatment (at least until the third phase is complete) needs to be
considered. One possible strategy would be to transition the probationer

into the third phase of treatment prior to leaving residential probation. This

156

could be accomplished by formally discharging the client from treatment
and formally readmitting him into the third phase of out-patient treatment at
the provider site just prior to discharging him from residential probation. As
part of the discharge planning process, the probationer would be required
to go to the out-patient treatment site and re-enroll in treatment. The
probationer would then be transferred to standard probation with a
treatment “expectation” as opposed to a treatment “recommendation.”
Probation officers can then continue to monitor compliance with treatment

expectations.

Substance Abuse Ralapse

Substance abuse relapse is an important indicator of successful
outcomes. As previously discussed, alcohol and THC relapse was
examine separately from relapse due to other drugs. Consistent with the
literature, the length of time an offender remains in treatment is a
significant predictor for both alcohol/'1' HC and drug relapse. The length of
time offenders spend in jail prior to being released on probation is also
significant in predicting alcohol/THC relapse. Offenders who are
incarcerated longer are less likely to relapse. The effect of incarceration on
alcohol and TCH relapse is interesting and again, only speculation can be

offered to explain this relationship. Research has shown that physical

157

withdrawal from chronic alcohol abuse can last for months during what is
often referred to as protracted abstinence syndrome or post acute
withdrawal (Goldman, 1983; Fitzgerald, 1988; Gorski and Miller,1986;
Kinney and Leaton, 1995). This prolong withdrawal is induced by the
damage to the central nervous system caused by alcohol. Recovery from
central nervous damage usually requires 6 to 24 months. During this
period, withdrawal symptoms may continue and include cognitive and
memory disturbances, excessive anxiety, irritability, insomnia, depressive
symptoms and more. Post acute withdrawal symptoms may appear 2
weeks into abstinence, generally peak at 3 to 6 months and may last for
several months thereafter (Gorski and Miller, 1986; Kinney and Leaton,
1995). Research indicates that the symptoms of long-term withdrawal
associated with central nervous system damage is related to relapse
(McCrady and Smith, 1981; Porjesz and Begleiter, 1983). Similarly, Ohlms
(1983) argues that because THC is absorbed by the body’s fat cells and is
rid of very slowly, heavy THC users experience a build up of THC in the
brain’s fat cells, or neurons. Even after long periods of abstinence TCH
users are susceptible to “flashbacks” which may be caused from the
release of the built up THC from the neurons. These flashbacks are
described as, the experience of a “THC high” and are theorized as being

trigger mechanisms for relapse. In contrast, both the early withdrawal and

158

second phase withdrawal from cocaine and crack cocaine are known to be
much shorter. The acute withdrawal phase typically lasts only 4 days and
is characterized by some of the same neurological systems noted in
alcohol withdrawal (e.g., agitation, depression). During this brief period
hypersomnia, fatigue and exhaustion usually occurs. As these symptoms
diminish post withdrawal begins characterized by anhedonia (lacking in
interest or pleasure; apathy). However, this post withdrawal period is
described only to last up to 10 weeks (Herridge and Gold, 1988; Gawin and
Kleber, 1986).

The most vulnerable periods of time for alcohol and drug abuser to
relapse is during both the acute and post acute withdrawal periods.
Clearly, alcohol and THC abusers experience the longest periods of post
acute withdrawal compared to other drugs. Confinement during this period
of time restricts access to the offender’s drug of choice and may aid him
(begrudgingly) in managing through periods of time when he is most
susceptible to relapse. Thus, where post acute withdrawal periods are
known to be long, longer periods of confinement may be associated with
reductions in relapse.

Age is also shown to be related to alcohol and THC relapse. The
older the offender the less likely they are to relapse. This finding is

consistent with previous research on aging offenders and criminal behavior

159

that show as offenders age criminal activity decreases and often ceases
completely (Inivin, 1970; Meisenhelder, 1977; Petersilia,et al., 1978;
Shover, 1985). These studies conclude that as offenders age or “mature”
they experience a number of changes which cause them to reduce their
criminal behavior. Among these changes are the development of new
commitments, a growing fear of arrest and incarceration, changes in self-
concept, motivations and approaches to problem solving. This maturation

process has also been used to explain recovery from heroin addiction.

 

However, results from these studies are conflicting. For example, using
data from the Federal Bureau of Narcotics registry, Winick (1962; 1964)
noted that most opiate addicts began use in their late teens and early
twenties and disappeared from the narcotics registry after age 35. Winick
therefore hypothesized that by age 35 most opiate addicts “mature out” of
the problems that led them to heroin use. Moreover, in studies of “natural
recovery” from heroin addiction31 researchers concluded that the ability to
quit heroin use was not necessarily related to treatment. Rather the
concept of “maturing out” preconditions recovery (Waldorf and Biemacki,
1981; Waldorf, 1983). However, subsequent longitudinal studies do not

support the maturation hypothesis to explain recovery from heroin

 

31

In contrast to recovery assisted by a formalized treatment process, the concept of natural
recovery is described as the ability to quit opiate use and maintain abstinence without treatment.

160

addiction. In two twenty year follow-up studies only 2 to 35 percent of
recovering heroin addicts could be classified as “matured out” by age 40
(Harrington and Cox, 1979; Vaillant, 1973).

These conflicting findings are difficult to explain. Moreover, in view
of the growing body of more recent scientific evidence suggesting a
possible inherited biological and neurological basis for some addictions, it
is difficult to conceptualize the ability of a person to “mature out” of a
genetic or biological anomaly. However, the concept of “drug dependence
syndrome” (DDS) may offer one possible explanation for the conflicting
findings. As a theory of relapse, the DDS postulates that relapse is a direct
function of the severity of the dependence BEFORE abstinence was begun
(Babor, et al., 1986). It is possible that there were major differences
between the severity of drug use which could range from drug misuse,
dependence to intense physical addiction. And the ability to “mature out” of
drug use may be directly related to the degree and severity of drug
dependence. Therefore the maturation process or “maturing out” theory
may be useful to explain drug abstinence based on both a person’s degree
of dependence or addiction and possibly their drug of choice. This
hypothesized relationship between the ability to “mature out” of drug abuse
and the degree of dependence and/or addiction may also explain why age

was found in this study to be a predictor of relapse among alcohol and THC

161

abusers relapse but not found to be associated with relapse for cocaine
and poly-drug abusers. However, data on the level of drug dependence or
addiction severity were not available for offenders in study sample.
Therefore it was not possible to test this assumption but further
investigation of this issue is indicated.

At the bivariant level, the proportion of offenders remaining alcohol
and THC free is slightly greater among offenders sentenced to residential
probation during the first few months, but observed differences diminish at
about one year. Although not as pronounced, a similar pattern is noted for
drug relapse. At the multivariant level these differences are not statistically
significant. However, a qualifying point needs to be made. The sample
size for this particular analysis was small with a large proportion of
censored observations. Comparisons of the risk ratios indicate that
observed changes are in the direction hypothesized. Given the consistent
pattern in the two analyses, there is reason to believe a larger sample
would demonstrate greater differences in the relative risk ratios between
cohorts.

On a practical level, the observed differences should not be ignored
by the programs. The data show that the proportion of offenders who
remain alcohol and drug during the first year of follow-up is slightly greater

for those offenders sentenced to residential probation and the relative risk

162

of relapse over time is slightly less. Even if outcomes are consider
equivalent between probation models, the positive treatment effects
observed in the residential probation cohort are accomplished with 6 to 10
weeks less treatment time. Given the significant relationship between time
in treatment and relapse, if offenders in residential probation where
continued in treatment after discharge there is reason to believe that
positive treatment outcomes would be greater among offenders sentenced
to residential probation compared to other probation models. Further

investigation of this issue is indicated.

Bacidiyism

Recidivism rates are always the greatest concern among officials
and policy makers and was the most problematic to evaluate in this study.
First, an overall failure model was constructed to evaluate the length of
time offenders remained on community supervision before receiving any
type of court sanction that resulted in jail or prison terms. The length of
time offenders remained in treatment is seen as the single best predictor of
failure, followed by offence category and drug of choice. Offenders
sentenced for drug crimes and offenders reporting their drug of choice as
either cocaine/crack or poly-drugs are more likely to fail. Next, failure due

to technical violation and new felony crimes are examined separately. In

163

both analyses, time in treatment remains the best predictor of failure.
However, the offence and drug of choice categories are not shown as
predictors of technical violations, as observed in the overall failure model.
What is significant in predicting technical violations is the probation model.
Specifically, the likelihood of failure due to technical violations is greatest
among offenders sentenced to the most intensive probation model
(residential probation with tether) and least among offenders among
offenders sentenced to the least intensive probation model. This is
consistent with previous research that suggests when monitoring
procedures are intensified the likelihood that probation violations will be
intensified is greater. The technical violation analysis strongly supports this
claim. However, an observation that was made while examining this
analysis also needs to be considered when interpreting the results.

The study sample included all offenders who met the eligibility
criteria for residential probation within a specified time period. In theory, all
offenders could have been sentenced to residential probation. The largest
proportion (56%) of technical violators in the study sample is seen in the
residential probation programs. Offenders sentence to the residential
probation programs (SPCC and SPCC/PES) are over 2 times more likely to
be technical violators compared to all other offenders. Compared to

offenders in standard probation, they are over 6 times more likely to be

164

sentence to residential probation for a technical violations. This is
interesting considering that the residential probation provider has a
program separate from SPCC or SPCC/PES specifically for technical
violators. This program is referred to as the Probation Rule Violator
Program for Circuit Court (PRV). PRV offenders were not part of this
evaluation. It appears that the judicial process also favors residential
probation for technical violations. Thus, a significant proportion of the
residential probation population are already technical violators. Because it
is assumed that the probability of being court sanction for a subsequent
technical violation is greater among probationers who have already been
sanctioned for a technical violation, it could be argued that this may also
explain why the failure rate for technical violations is the greatest among
offenders in the residential probation cohorts. A more appropriate analysis
of residential probation may be to examine the outcomes of offenders
sentenced on probation violations only. This would increase the sample,
eliminate offenders sentenced to standard probation and improve the ability
to detect true differences.

The large proportion of technical failure was problematic when
measuring the effect of residential probation on felony crime (A complete
discussion of the issues are found on pp. 118 -119). The conclusion is that

inferences regarding the effect of the various probation models can not be

165

 

drawn. This also eliminated the ability to test hypothesis 4: “Of the
probationers rearrested for new crimes, the severity of the new offence will
be less for offenders sentenced to residential probation”. However, the
analysis of felony crime failure did produce meaningful information. Even
with a limited sample in this analysis, length of time in treatment is shown
to be the single best prediction of recidivism. Also related to felony crime
is education level; the higher the education the less likely the offender is to
recidivate. This is consistence with the literature and lends support to the
residential probation programs in providing educational opportunities on-
slte. Offence category and drug of choice are shown to have a significant
effect on recidivism as well. Offenders sentenced for drug crimes are over
3 times more likely to commit new felony crimes compared to offenders
sentenced for violent crimes. Offenders who report their drug of choice to
be cocaine/crack or poly-drugs are over 2 times more likely to be
rearrested for a new felony crimes. The above substantiates what is
already known to providers in this jurisdiction: the population most at risk
are offenders with limited education, whose drug of choice is other than
alcohol or THC, and are involved in felony drug crimes.

Although this evaluation documents much of which is already known
in the literature, the information and recommendations for change are

specific to a particular population and jurisdiction. The information is

166

intended to be use by officials to enhance the delivery system in efforts to
improve outcomes where possible. Most importantly, length of time in
substance abuse treatment is proven to be significant in effecting all
outcomes in this jurisdiction. Treatment is working here, and residential
probation programing is the most effective in ensuring compliance with

court mandated treatment.

Direction for Future Research:

In response to the many issues identified in the probation literature,
the primary purpose of this research was to examine the impact of a
uniquely structured residential ISP program to determine if a different
combination of treatment and surveillance could produce different
outcomes for a specific population of drug involved felony offenders. While
several findings in this study mirror what has been previously documented
in the probation literature, other findings suggest the need to modify and
retest program models in efforts to refine existing programs and improve
the effectiveness of probation. For example, although results in this study
document that drug involved felony offenders sentenced to residential
probation are more likely to become engaged in the treatment process, the
structure of both the residential probation and substance abuse treatment

programs impact the length of time offenders are expected to remain in

167

treatment. Specifically, in some cases, the length of time offenders who
are sentenced to residential probation are expected to remain in treatment
is notably less compared to some offenders sentenced to other probation
programs. However, even with less treatment time, alcohol and drug
relapse rates among offenders sentenced to residential probation are not
greater. There is evidence to suggest that a small adjustment in the
program’s policies could ensure that offenders would remain in treatment
longer and ultimately produce better outcomes compared to other
probation programs. Following this refinement, a retest of the program’s
effectiveness is indicated which could provide information for improving the
effectiveness of probation.

Although the residential probation program evaluated in this study
was not intentionally designed to primarily service offenders sentenced on
technical violations, the data indicate that a greater proportion of the
program’s participants are in fact sentenced to the program for a technical
violation while on probation for a felony offence. Given this reality, it is
suggested that a more appropriate evaluation of this residential probation
would be to examine the outcomes of offenders sentenced on technical
violations between the various probation models. This subsequent study
would help to further determine whether a different combination of

treatment and surveillance will produce different results in different

168

populations.

Based on the findings in this study, it is also suggested that studies
should be designed to examine the effects of incarceration on offenders’
motivation for treatment. These studies could help determine if optimal
levels of incarceration exist for different offenders under various conditions,
which could ultimately serve as a valuable treatment tool. In addition,
longitudinal studies which clarify the relationship between age and drug
relapse, not excluding the possible interaction of addiction severity, are
indicated. Such studies would allow researchers to examine more closely
the processes of relapse and recovery.

In summary, this research clearly demonstrates that the substance
abuse treatment oriented residential probation program examined is
significantly more effective in ensuring that drug involved felony offenders
comply with court mandated substance abuse treatment compared to other
types of probation programs. Further, even though the expected length of
time probationers sentenced to residential probation are to spend in
treatment is approximately 6 to 10 weeks less compared to probationers
sentenced to other probation programs, they are somewhat less likely to
relapse due to alcohol and drug use over time. This provides evidence
suggesting that significantly improved outcomes could be achieved among

probationers sentenced to residential probation with a minor modification in

169

the residential probation program’s policy that ensures these offenders
would be retained in treatment longer once the probationer is discharged
from residential probation.

Moreover, this research documents that the longer offenders are
incarcerated prior to being released on probation the less likely they are to
enter substance abuse treatment. Most program policies tend to support
punishment of the offender as opposed to treating the drug-involved
offender as having a chronic relapsing diseased characterized by “drug
seeking behavior.” Given the significant relationship between substance
abuse treatment and positive outcomes, the problems with significant jail
overcrowding and the documented effectiveness of the residential
probation program in ensuring compliance with court mandated treatment,
this study suggests that residential probation many be a more cost effective
and efficient sanctioning strategy compared to long periods of incarceration
in combination with other probation program models.

Although it is known that intensive monitoring or closer supervision
have little impact on subsequent criminal behavior, this study documents
that more intensive monitoring does affect whether offenders comply with
treatment mandates, which in turn affects subsequent behavior. These
findings appear congruent with deterrence theory in that it is assumed that

the “perceived threat” of punishment and the “certainty of punishment” for

170

 

 

failing to comply with court mandated treatment are greater for offenders
sentenced to residential probation. However, further testing of this
assumption is indicated.

Overall, this research supports the claim that to improve positive
outcomes intensive supervision programs need to change the way in which
they are intensive by shifting the emphasis away from incapacitation and
punishment toward a more integrated approach of intervention and
substance abuse treatment. The “intensive” focus needs to be on intensify
the monitoring of treatment compliance. To move forward, future probation
program models that fully incorporate this change in focus need to be

developed and tested.

171

 

BIBLIOGRAPHY

 

172

VII. BIBLIOGRAPHY

Abadinsky, Howard (1989). Dnrg Abuse: An Introduction. Chicago, IL:
Nelson-Hall.

Abbott, M.W., Gregson, RAM. (1981). “Cognitive dysfunction in the
prediction of relapse in alcoholics.” Joumal of Studies on Alcohol
422230 - 243.

Anglin, MD. (1988). "The efficacy of civil commitment in treating narcotic
addiction." In: Compulsory Treatment of Drug Abuse: Research and
Clinical Practice. National Institute on Drug Abuse: Research
Monograph 86. Washington, DC: US. Government Printing Office:
8 - 33.

Anglin, MD, and Hser, Y. (1990a). "Treatment of drug abuse.” Pub. No.
NCJ-125249 In: Drugs and Crime. edited by, M.Tonry and J.Q.
Wilson, Chicago: University of Chicago Press. 13: 393-460.

Anglin, MD, and Hser, Y. (1990b). "Legal coercion and drug abuse
treatment: research findings and social policy implications.” In: Hand
Book of Drug Control in the U. 8. edited by, A. lnciardi, Pub. No.
NCJ-126327; West Port, CT: Greenwood Press: 151-176.

Austin, J., Baird, C., Clear, T., DeComo, R., Holien, D., Neuenfeldt, D.,
Tupper, S., Wagner, D., and Wiebush, R. (1990). Intensive
Supervision in the United States; A review of Recent Experience and
Critical Issues with Recommendations for Future Development.
Madison, WI: National Council on Crime and Delinquency.

Babor, T.F.; Cooney, ML; and Lauerman, R.J (1986). “The drug
dependence syndrome concept as an organizing prediction of
relapse.” In: Relapse and Recovery in Dnrg Abuse, edited by Tims,
FM; and Leukefeld, C.G., National Institute on Drug Abuse
Research Monograph 72. Washington, DC: US. Government
Printing Office: 20 - 35.

173

Banks, J., and Rardin, R.L. (1977). Phase 1; Investigation of Intensive
Special Probation. National Evaluation Program Submitted to the
National Conference on Criminal Justice Evaluation; Pub. No.
NCJ - 054062. Washington, DC.

Barr H, and Antes, D. (1981 ). Factors Related to Recovery and Relapse in
Follow-Up. Final Report of Project Activities: Grant No. H81-
DAO1864. Rockville, MD: National Institute on Drug Abuse.

Begleiter, H. (1981). “Brain dysfunction and alcoholism: Problems and
prospects.” Alcoholism: Clinical and Experimental Research 5(2):
264 - 266.

Brecher, EM. (1972). Licit and Illicit Drugs. Boston: Little, Brown, and Co.

Bureau of Justice Assistance (1988). "Intensive supervision probation and
parole (IPS).” Program Brief. Pub. No. NCJ-10663. Washington, DC:
US Department of Justice.

Bureau of Justice Statistics (1992a). Source Book of Statistics, 1991. US.
Department of Justice, Washington, DC: Government Printing Office.

Bureau of Justice Statistics (1992b) "Drug Enforcement and Treatment in
Prison, 1990." Special Report. US. Department of Justice, Office of
Justice Programs, Pub. No. NCJ-134724. Washington, DC:
Government Printing Office.

Bureau of Justice Statistics (1996a). Correctional Populations in the United
States, 1995. US Department of Justice, Office of Justice
Programs, Pub. No. NCJ-163916. Washington, DC: Government
Printing Office.

Bureau of Justice Statistics (1996b). Prison and Jail Inmates, 1995. US.
Department of Justice, Office of Justice Programs, Pub. No. NCJ-
161132. Washington, DC: Government Printing Office.

Bureau of Justice Statistics (1997a). Prison and Jail Inmates, at Midyear

1996. US. Department of Justice, Office of Justice Programs, Pub.
No. NCJ-162843. Washington, DC: Government Printing Office.

174

Bureau of Justice Statistics (1997b). Characteristics of Adults on
Probation, 1995. US. Department of Justice, Office of Justice
Programs, Pub. No. NCJ-164267. Washington, DC: Government
Printing Office.

Bureau of Justice Statistics (1998). Prison and Jail Inmates, at Midyear
1997. US. Department of Justice, Office of Justice Programs, Pub.
No. NCJ-167247. Washington, DC: Government Printing Office.

Byme, J.M. (1986). “The control controversy: A preliminary examination
of intensive probation supervision programs in the United States.”
Federal Probation 50: 4-16, (NCJ-111579).

Byme, J.M. (1990). “The future of intensive probation supervision and the
new intermediate sanctions.” Crime and Delinquency 36: 6-41.

Byme, J.M.; Lurigio, A.J.; and Baird, C. (1989). "The effectiveness of the
new intensive supervision programs.” Research in Corrections
2:(2):1-48. (NCJ-121077).

Clear, Todd; Braga, Anthony A. (1995). "Community corrections." In:
Crime. edited by; James O. Wilson and Joan Petersilia, San
Francisco: Institute for Contemporary Studies.

Clear, Todd; Cole, George (1994). American Corrections: Third Ed.
Pacific Grove, California: Brooks/Cole Publishing Co.

Clear, T.R.; Hardyman, PL. (1990). “The new intensive supervision
movement." Crime and Delinquency 36(1): 42-60.

Cloninger, C.R.; Bohman, M.; Sigvardsson, S. (1981). “Inheritance of
alcohol abuse: Cross-fostering analysis of adopted men.” Archives of
General Psychiatry 382861 - 868.

Cohen, Gerald. (1978). “The synaptic properties of some
tertahydrosioquinoline alkaloids. Alcoholism: Clinical and
Experimental Research 2(2) 121-132.

Collins, J.J.; Allison, M. (1983). “Legal coercion and retention in drug abuse
treatment.” Hosp Community Psychiatry 34(12):1145 - 1149.

175

 

 

Cook, Foster (1992). "TASK: Case management models linking criminal
justice treatment." In: Progress and Issues in Case Management,
edited by Ashery, R.S. National Institute on Drug Abuse Research

Monograph 127. Washington, DC: US Government Printing Office:
368-382.

Cotton, NS. (1979). “The familial incidence of alcoholism: A review.”
Journal of Studies on Alcoholism 40:89 - 116.

Cox, DR. (1972). “Regression models and life tables.” J R Stat Soc B 34:
187 - 220.

Cuffin, MA, and Shilton, MK. (1991). Variation on Felony Probation:
Persons Under Supervision in 32 Urban Suburban Counties. US.

Department of Justice, Bureau of Justice Statistics, Washington, DC:
US. Government Printing Office.

De Leon, George (1984). “Program-based evaluation research in
therapeutic communities.” In: Drug Abuse Treatment Evaluation:
Strategies, Progress and Prospects edited by Tims, F.M.; Ludford,

J.P. National Institute on Drug Abuse Research Monograph 51.
Washington, DC: US Government Printing Office.

De Leon, George (1988). "Legal pressure in therapeutic communities."
In: Compulsory Treatment of Drug Abuse: Research and Clinical
Practice, edited by Leukefeld, C.G.; Tims, F .M. National Institute on

Drug Abuse: Research Monograph 86. Washington, DC: US.
Government Printing Office: 160-177.

De Soto, C. B.; O’Donnell, W.E.; Alfred, L.J.; Lopes, CE. (1985).
“Symptomatology in alcoholics at various stages of abstinence.”
Alcoholism: Clinical and Experimental Research 9(6):505 - 512.

Erwin, BS. (1986). “Turning up the heat on probation in Georgia.” Federal
Probation 50(2): 7-24. (NCJ-111580).

Erwin, BS. (1987). New Dimension in Probation: Georgia’s Experience
with Intensive Probation Supervision (IPS). Research in Brief. Bureau
of Justice Statistics, National Institute of Justice, Pub. No. NCJ-
102848. Washington, DC: US. Government Printing Office.

176

Erwin, BS. (1990). “Old and new tools for the modern probation officer.”
Crime and Delinquency 36(1): 61 -74.

Erwin, BS, and Bennett, LA. (1987). New Dimensions in Probation:
Georgia’s Experience with Intensive Probation Supen/ision (IPS).
Research in Brief. Department of Justice, National Institute of
Justice, Pub. No. NCJ-102848. Washington, DC: US. Government
Printing Office.

Field, G., (1989). "The effects of intensive treatment on reducing the
criminal recidivism of addicted offenders." Federal Probation 53(1):
51-56.

Fitzgerald, K.W. (1993). Alcoholism: The Genetic Inheritance. Lake Forest,
IL: Whales’ Tale Press.

 

Fulton, B. and Gendreau, P. (1995). “APPA’s prototypical intensive
supervision program: ISP as it was meant to be.” Perspectives
(Spring): 25-41.

Gawin,F.H., and Kleber, H.D. (1986). “Abstinence symptomatology and
psychiatric diagnosis in cocaine abusers.” Archives of General
Psychiatry 43(2): 1 07 - 1 13.

Geerken, MR, and Hayes, H.D. (1993). "Probation and parole: Public
risk and the future of incarceration alternative.” Criminology 31 (4):
549-564.

Gennaro, Vito F. (1987). “Felony probation and recidivism: Replication
and response.” Federal Probation 50(17).

Gerstein, D., Johnson, R.A., Harwood, H.J., Fountain, D., Suter, N., and
Malloy, K. (1994). Evaluating Recovery Services: The California
Drug and Alcohol Assessment. Pub. No. NCJ-157812. Sacramento:
State of California, Department of Alcohol and Drug Programs.

Gold, M.S.; Washington, A.M.; Dackis, CA. (1985). ”Cocaine abuse:
Neurochemistry, phenomenology, and treatment.” In: Cocaine Use in
America: Epidemiologic and Clinical Perspectives, edited by Kozel,
N.J.; Adams, E.H. Rockville, MD: National Institute on Drug Abuse:
130 - 150.

 

177

Goldman, MS. (1983). “Cognitive impairment in chronic alcoholics: Some
cause for optimism.” Am Psychologist 38:1045 - 1054.

Goldstein, A. (1978). “Endorphins: Physiology and clinical implications.”
Ann NYAcad Sci 31 1: 49 - 58.

Goodwin, D.W. (1980). “Genetics of alcoholism, substance and alcohol
actions/misuse.” Clinical Science Review 1:101 - 117.

Gorski, Terence T; Miller, Merlene (1986). Staying Sober: A Guide for
Relapse Prevention. Independence, MO: Independence Press.

Gostin, Lawewnce O. (1991). “Compulsory treatment for drug-dependent
persons: Justification for a public health approach to drug
dependencey.” The Milbanks Quarterly 69(4).

Guynes, R. (1988). Difficult Clients, Large Caseloads Plague Probation,
Parole Agencies.” Research in Action. Department of Justice,
National Justice Institute, Pub. No. NCJ-113768. Washington, DC:
US. Government Printing Office.

Hamiliton, Murray C., Blum, Kenneth and Hirst, Maurice (1978).
“Identification of an isoquinline alkaliod after chronic exposure to
ethanol.” Alcoholism: Clinical and Experimental Research 2 (2) 133-
142.

Harrington, P., and Cox, TA (1979). “A twenty-year follow-up of narcotic
addicts in Tucson, Arizona.” Am J Drug Alcohol Abuse 6:25 - 37.

Herridge, P. and Gold, M. (1988). “Pharmacological adjuncts in the
treatment of opioid and cocaine addicts.” Journal of Psychoactive
Drugs 20(3): 233 - 242.

Ho, W.K.K.; Wen H.L.; Ling, N. (1980). “Beta-endorphins-Iike
immunoactivity in the plasma of herion addicts and normal subjects.”
Neuropharmacology 19:117 - 120.

Holland, S. (1983).“Evaluating community based treatment programs: A

model for strengthening inferences about effectiveness.”
International Journal of Therapeutic Communities 4(4): 285-306.

178

Hubbard, R.L., Collins, J.J., Rachal, JV, and Cavanaugh, ER. (1988).
“The criminal justice client in drug abuse treatment.” In: Compulsory
Treatment of Drug Abuse: Research and Practice, edited by
Leukefeld, C.G.; Tims, F.M. National Institute on Drug Abuse
Research Monograph 86. Washington, DC: US. Government
Printing Office: 57-80.

lnciardi, James A. (1988). "Some considerations on the clinical efficacy of
compulsory treatment: Reviewing the New York experience." In:
Compulsory Treatment of Drug Abuse: Research and Clinical
Practice, editied by Leukefeld, C.G.; Tims, F.M. National Institute on
Drug Abuse Research Monograph 86. Washington, DC: US.
Government Printing Office: 126-137.

Irwin, John ( 1970). The Felon. Englewood Cliffs, NJ: Prentice-Hall.

Irwin, J., and Austin, J. (1987). It’s About Time: Solving America’s Prison
Crowding Problem. National Council on Crime and Delinquency: San
Francisco, CA.

Kay, R. (1977). “Proportional hazard regression models and the analysis of
censored survival data.” Applied Statistics 34:227 - 337.

Kinney, J., and Leaton, G. (1995). Loosening the Grip: A Handbook of
Alcohol Information. Fifth Edition, St. Louis: Times Mirror/Mosby
College Publishing.

Khantzian, Edward J. (1985). “The self-medication hypothesis of addictive
disorders: Focus on heroin and Cocaine Dependence.” American
Journal of Psychiatry 142 :1259 - 64.

Klein, Andrew R. (1997). Altemative Sentencing, Intermediate Sanctions,
and Probation. Cincinnati: Anderson.

Langan, Patrick A. (1994). "Between prison and probation: Intermediate
sanctions." Science 264:791-93.

Langan, PA, and MA. Cunniff. (1992). Recidivism of Felons on
Probation, 1986-89. Special Report. Department of Justice, Bureau
of Justice Statistics, Pub. No. NCJ-134177. Washington, DC: US.
Government Printing Office.

179

Langenauer, B.J., and Bowden, CL. (1971). "A follow-up study of narcotic
addicts in the NARA program." American Journal of Psychiatry 128:
41-46.

Latessa, Edward J. (1986). “The cost effectiveness of intensive
probation.” Federal Probation 50(2): 70-74.

Leshner, Alan I. (1997). “Addiction is a brain disease, and it matters.”
Science 278: 45-47.

Leukefeld, Carl G., and Tims, Frank M. (1988a). "An introduction to
compulsory treatment for drug abuse: Clinical practice and
research." In: Compulsory Treatment of Drug Abuse: Research and
Clinical Practice, edited by Leukefeld, C.G.: Tims, F.M. National
Institute on Drug Abuse Research Monograph 86. Washington, DC:
US. Government Printing Office: 1-7.

Leukefeld, Carl G., and Tims, Frank M. (1988b). "Compulsory Treatment: A
review of findings." In: Compulsory Treatment of Drug Abuse:
Research and Clinical Practice, edited by Leukefeld, C.G.: Tims,

F .M. National Institute on Drug Abuse Research Monograph 86.
Washington, DC: US. Government Printing Office: 236-252.

Levinthal, Charles F. (1988). Messengers of Paradise: Opiates and the
Brain. Garden City, NY: Doubleday.

Lilienfield, D., and Stolley, P. (1994). Foundations of Epidemology. Third
Edition. New York: Oxford Press Inc.

Lipton, Douglas S. (1995). The Effectiveness of Treatment for Drug
Abusers Under Criminal Justice Supervision. US. Department of
Justice, National Institute of Justice, Pub. No. NCJ-157642.
Washington, DC: US. Government Printing Office.

Lurigio, A.J. (1997). A One Year Statewide Evaluation of IIIinois's
Intensive Probation Supervision and Intensive Dnrg Abuser
Probation Programs. Final Report, Loyola University, Chicago, IL.

Lurigio, A.J. (1990). “Introduction.” Crime and Delinquency 36 (1 )23-5.

Martin, Joseph C. (1982). No Laughing Matter: Chalk Talks on Alcohol.
San Francisco, CA: Harper and Row, Publishers.

180

Maquire, K., and Flanagan, T.J. (1991). Source book of Criminal Justice
Statistics; 1991. Washington, DC: US. Department of Justice,
Bureau of Justice Statistics.

Maddux, J.F. (1987). "History of the hospital treatment programs, 1935-
74." In: Drug Addiction and the U. S. Public Health Service, edited by
Martin, W.R. and lsbell. National Institute on Drug Abuse. DHHS
Pub. No. 77-434. Washington, DC: US. Government Printing Office:
217-250.

Maddux, J.F. (1988). "Clinical history with civil commitment." In:
Compulsory Treatment of Drug Abuse: Research and Clinical
Practice, edited by Leukefeld, C.G.; Tims, F .M. National Institute on
Drug Abuse Research Monograph Series, 86; Washington, DC: US.
Government Printing Office: 62-83.

McCrady, B.S.; Smith, DE. (1981). “Implications of cognitive impairment
for the treatment of alcoholism.” Alcoholism: Clinical and
Experimental Research 10(2).

McGlothlin, W.H., Anglin, M.D., Wilson, B.D.(1977a). An Evaluation of the
California Civil Addicts Program. National Institute on Drug Abuse
Services Research Monograph Series. DHEW Pub. No.(ADM) 78-
558. Washington, DC: Supt. Of Docs., US. Govt. Printing Office.

McGlothlin, W.H., Anglin, M.D., Wilson, B.D.(1977b). “A follow-up of
admissions to the California civil addiction program.” Am J Drug
Alcohol Abuse 4 (2):197-199.

McGlothlin, W.H., Anglin, M.D., Wilson, B.D.(1978) "Narcotic addiction
and crime." Criminology 16 (Nov.):293—315.

Meisenhelder, Thomas N. (1977). “An exploratory study of existing from
criminal careers.” Criminology 15 (Nov.): 319 - 334.

Mohr, Lawrence B. (1988). Impact Analysis For Program Evaluation.
Chicago, IL: The Dorsey Press.

Morgan, Howard W. (1981).-Drugs in America: A Social History, 1880-
1980. Syracuse, NY: Syracuse University Press.

181

 

 

Morgan, Katheryn. (1993). “Factors influencing probation outcome: A
review of the literature.” Federal Probation 57(2): 23—29.

Motulsky, Harvey (1995). Intuitive Biostatistics. Oxford, NY: Oxford
University Press.

Myers, RD. (1978). “Tetrahydroisoquinolines in the brain: The basis of an
animal model of alcoholism. Alcoholism: Clinical and Experimental
Research 2 (2).

Ohlms, David L. (1993). The Disease Concept of Alcoholism. Gary
Whieaker Corporation; Cahokia, IL.

Ohlms, David L. (1983). Pot. Gary Whieaker Corporation; Cahokia, IL.

Pearson, Frank S. and Bibel, Daniel B. (1986). “New Jersey’s intensive
supervision program: What it is like? How it is working?” Federal
Probation 50(2): 22-31.

Pearson, Frank S.; Harper, Alice G. (1990). “Contingent intermediate
sentences: New Jersey’s Intensive Supervision Program.” Crime
and Delinquency 36: 75-86.

Petersilia, Joan. (1990). “Conditions that permit intensive supervision
programs to survive.” Crime and Delinquency 36 (1): 126-145.

Petersilia, Joan. (1993). "Measuring the performance of community
corrections." In: Performance Measures for the Criminal Justice
System. US. Department of Justice, Bureau of Justice Statistics,
Washington, DC: US. Government Printing Office.

Petersilia, Joan (1995). “A crime control rationale for reinvesting in
community corrections”. Prison Journal 75(4): 479-496.

Petersilia, Joan (1997). "Probation in the United States." In: Crime and
Justice; A Review of Research Vol. 22, edited by, M. Tonry, Chicago:
The University of Chicago Press.

Petersilia, J., Greenwood, P.W., and Lavin, M. (1978). Criminal Careers of

Habitual Felons. Washington, DC: National Justice Institute. NCJ#
45351 .

182

Petersilia, J., and Turner, S. (1990). "Comparing intensive and regular
supervision for high-risk probationers: Early results from an
experiment in California." Crime and Delinquency 36287-111.

Petersilia, J., and Turner, S. (1992). "An Evaluation of Intensive Probation
in California." The Joumal of Criminal Law and Criminology 82(5):
610-658.

Petersilia, J., and Turner, S. (1993a). Evaluating Intensive Supervision
Probation/Parole: Results of a Nationwide Experiment. Research in
Brief. National Institute of Justice, Department of Justice, Pub. No.
NCJ-141637. Washington, DC: US. Government Printing Office.

Petersilia, J., and Turner, S. (1993b). "Intensive probation and parole." In:
Crime and Justice: A Review of the Research Vol. 17. edited by,
Michael Tonry. Chicago: University of Chicago Press.

 

Porjesz, B.: Begleiter, H. (1983). “Brain dysfunction and alcohol.” In: The
pathogenesis of Alcoholism — Biological Factors, edited by Kissin,
B.; Begleiter, H. New York: Plenum Press: 415 -483.

Prichard, David A. (1979). “Stable predictors of recidivism: A summary.”
Criminology. 17(15).

Robinson v. California, 370 US. 660, 82 S.Ct. 1417, 8 L.Ed.2d 758 (1962).
In: Krantz, Sheldon, The Law of Corrections and Prisoner's Rights,
Third Ed.,St. Paul Minn: West Publishing Co., 1986.

Rossi, PH. and Freeman, HE. (1982). Evaluation; A Systematic Approach.
Second Edition. Beverly Hills, CA: Sage Publications, Inc.

 

SAS Institute Inc. (1997). SAS/Software: Changes and Enhancements
Through Release 6.12. Cary, NC: SAS Institute INC: 413 - 431.

Schuckit, Marc A, (1980). “A theory of and drug abuse: A genetic
approach.” In: Theories on Drug Abuse: Selected Contemporary
Perspectives. Edited by Lettieri , D.C.; Sayers, M.; and Pearson,
H.W. National Institute on Drug Abuse Research Monograph 30.
Washington, DC: US Government Printing Office.

183

Schuckit, Marc A.; Rayses, V. (1979). “Ethanol ingestion: Differences in
Blood acetaldehyde concentration in relatives of alcoholics and
controls.” Science 203.

Schuckit, Marc A.; Li Ting Kai; Clonninger, C.; Deitrich, C.R.; Richard, A.
(1985). “The genetics of alcoholism.” Alcoholism: Clinical and
Experimental Research 9(6):475 - 492.

Shover, Neal (1985). Aging Criminals. Beverly Hills, CA:Sage Publications.

Simpson, DD, and Friend, H.J. (1988). "Legal status and long-term
outcomes for addicts in the DARP follow-up project." In: Compulsory
Treatment of Drug Abuse: Research and Clinical Practice, edited by
Leukefeld, C.G.; Tims, F.M. National Institute on Drug Abuse
Research Monograph 86. Washington, DC: US. Government
Printing Office: 81-97.

Simpson, DD, and Marsh, KL. (1986). “Relapse and recovery among
opioid addicts 12 years after treatment.” In: Relapse and Recovery in
Drug Abuse, edited by Tims, F.M.; and Leukefeld. National Institute
on Drug Abuse Research Monograph 72. Washington, DC: US.
Government Printing Office: 86 -103.

Simpson, DD, and Sells, SB. (1982). "Effectiveness of treatment for
drug abuse: An overview of the DARP research program." Adv.
Alcohol and Substance Abuse 2(1): 7-29.

Stephens, R., and Cottrell, E. (1972). "A follow-up study of 200 narcotic
addicts committed for treatment under narcotic addict rehabilitation
act (NARA)." Br J Addict 67: 45-53 .

Thompson, Douglas R. (1987). Intensive Probation Supervision in Illinois:
A First Year Progress Report on the Evaluation. Chicago: Center for
Research in Law and Justice, University of Illinois at Chicago.

Thompson, Douglas R. (1990). “How plea bargaining shapes intensive

probation supervision policy goals.” Crime and Delinquency 36(1):
146-161.

184

 

 

Tims, F .M.; Fletcher, B.W.; Hubbard, R.L. (1991). “Treatment outcomes for
drug abuse clients.” In: Improving Drug Treatment, edited by
Pickens, R.W: Leukefeld, C.G.; Schuster, C.R. National Institute on
Drug Abuse Research Monograph 106. Washington, DC: US.
Government Printing Office: 93 -113.

Tims, Frank M. and Carl G. Leukefeld. (1992). "The challenges of drug
abuse treatment in prisons and jails." In: Drug Abuse Treatment in
Prisons and Jails, edited by Leukefeld, C.G.; Tims, F.M. National
Institute on Drug Abuse Research Monograph 118. Washington, DC:
US. Government Printing Office: 1-7.

Tittle, CR. (1980). “Evaluating the deterrent effects of criminal sanctions.”
In: Handbook of Criminal Justice Evaluation, edited by, Klein, M.W.;
Teilmann, K.S.. Beverly Hill, CA: Sage Publications.

Tonry, Michael. (1990). “Stated and latent functions of IPS.” Crime and
Delinquency 36( 1 ):1 1 2-125.

Tonry, Michael, and Mary Lynch. (1996). “Intermediate sanctions.” In:
Crime and Justice; A Review of Research Vol 20. edited by: Michael
Tonry. Chicago: University of Chicago Press.

Turner, 8., Petersilia, J., and Deschenes, ER (1992). "Evaluating
intensive supervision probation/parole (ISP) for drug offenders."
Crime and Delinquency 38(4):539-556.

US. General Accounting Office. (1990). Intermediate Sanctions: Their
Impacts on Prison Crowding, Costs, and Recidivism are Still Unclear.
No. GAO/PEMD-90-21. Washington, DC.

US. Cong. House. Committee on Narcotics and Abuse and Control.
(May, 1991). Hearing on: Drug Treatment in Prisons. 102nd Cong.,
1st sess. HR. 730. Washington, DC: US. Government Printing
Office.

Vaillant, GE. (1973). “A 20-year follow-up of New York’s narcotic addicts.”
Arch Gen Psychiatry 29:237 - 241.

Waldorf, D. (1973). “Natural recovery from opiate addiction: Some social-
psychological processes of untreated recovery.” J Drug Issues
29:237 - 241.

185

Waldorf, Dan and Biemacki, Patrick (1981).The natural recovery from
opiate addiction: some preliminary findings.” Journal of Drug Issues
11(Winter): 61 - 74.

Weiss, Carol H. (1972). Evaluation Research: Methods forAssessing
Program Effectiveness. Englewood Cliffs, New Jersey: Prentice-Hall,
Inc.

Wexler, H.K.; Lipton, D.S.; and Johnson BB. (1988). A Criminal Justice
System Strategy for Treating Cocaine-Heroin Abusing Offenders in
Custody. Issues and Practices in Criminal Justice. Department of
Justice, Pub. No. NCJ-108560. Washington, DC: US. Government
Printing Office.

Wexler, H.K.; Falkin, GP; and Lipton, D.S. (1990). "Outcome evaluation of
a prison therapeutic community for substance abuse treatment."
Criminal Justice Behavior 17: 71 -92.

Wexler, H.K. Falkin, G.P., Lipton, D.S., and Rosenblum, AB. (1992).
“Outcome evaluation of a prison therapeutic community for
substance abuse treatment.” In: Drug Abuse Treatment in Prisons
and Jails, edited by Leukefeld, C.G.; Tims, F .M. National Institute on
Drug Abuse Research Monograph 118. Washington, DC: US.
Government Printing Office: 156 - 174.

Winick, C. (1962). “Maturing out of narcotic addiction.” Bull Narc 14:1- 7.

Winick, C. (1994). “The life cycle of the narcotic addict and of addiction.”
Bull Narc 16:1-1 1.

Wright, Martin. (1991). Justice for Victims and Offenders.
Philadelphia: Open University Press.

186

 

General References

Alford, GS. (1980). "Alcoholics anonymous: An empirical outcome study."
Addictive Behavior. 5: 359-370.

Andreoni, James (1995). “Criminal deterrence in the reduced form: A new
perspective on Ehrlick’s seminal study.” Economic Inquiry 33(3):
476 -308

Anglin, M.D., (1982)."Alcohol and criminality." In: Encyclopedia Handbook
of Alcoholism, edited by Patterson E.M; Kaufman, E. New York:
Gardner: 383-394.

Anglin, MD, and Hser, Y. (1991). "Criminal justice and the drug-abusing
offender: Policy issues in coerced treatment." Behavior Science and
the Law 9: 46-62.

Baird, SC. and Wagner, D. (1990). “Measuring diversion: The Florida
community control program.” Crime and Delinquency 36(1):112-125.

Ball, J.C.; Rosen, L.; Flueck, SA; and Nurco, ON. (1981). "The criminality
of heroin addicts: When addicted and when off opiates." In: The
Drug-Crime Connection, edited by lnciardi, J.A. Beverly Hills, CA:
Sage: 39-65.

Byme, J.M.; Lurigio, A.J.; and Petersilia, J.M. (1992). Smart Sentencing:
The Emergence of Intermediate Sanctions. Newbury Park, CA:
Sage.

Califano, Joseph A. Jr. (1990). "Overview: Drug policy for the 1990's.” Yale
Law and Policy Review 82(1).

Clear, Todd R., and Shapiro, Carol (1986). “Identifying high risk
probationers for supervision in the community: The Oregon model.”
Federal Probation 50(2): 42-49.

Cochran, D., Corbett, R.P. Jr., and Byme, J.M.(1986). “Intensive probation

supervision in Massachusetts: A case study in change.” Federal
Probation 50(2): 32-41.

187

 

Cook, Foster L., and Weinman, Beth A., et al. (1988). "Treatment
alternatives to street crime." In: Compulsory Treatment of Drug
Abuse: Research and Clinical Practice, edited by Leukefeld, C.G.;
Tims, F.M. National Institute on Drug Abuse Research Monograph
86. Washington, DC: US. Government Printing Office, 99-105.

Cruze, A.M.; Harwood, H.J.; Kristiansen, P.L.; and Collins, J.J. (1981).
Economic Costs to Society of Alcohol and Drug Abuse and Mental
Illness. Research Triangle Park, NC.

Cullen, F.T.; and Gilbert, KB (1982). Reaffirming Rehabilitation.
Cincinnati, OH: Anderson.

Cullen, F.T.; Cullen, J.B.; and Wozniak, J.F. (1988). "Is rehabilitation
dead? The myth of the punitive public.” Journal of Criminal Justice
16:303.

De Leon, George (1991). "Retention in drug-free therapeutic
communities." In: Improving Drug Abuse Treatment, edited by
Pickens, R.W.; Leukefeld, C.G.; Schuster, C.R. National Institute on
Drug Abuse Research Monograph 106. Washington, DC: US.
Government Printing Office: 218-245.

Epstein, Edward J. (1977). Agency Fear: Opiates and Political Power in
America. New York: Putnam's.

Fein, Rashi (1984). Alcohol in America: The Price We Pay. Newport,
Beach California: Carelnstitute.

Field, G. (1984). "The Cornerstone program: A client outcome study.”
Federal Probation 49(2):50—55.

Goldberg, Raymond (1994). Drugs Across the Spectrum. Minneapolis/St.
Paul: West Publishing Co.

Goode, E. (1993). Drugs and American Society; 4th Edition. New York:
McGraw-Hill.

Gregrich, John (1992). "Management of the drug-abusing offender." In:
Drug Abuse Treatment in Prison and Jails, edited by Leukefeld, C.G.;
Tims, F .M. National Institute on Drug Abuse Research Monograph
118. Washington, DC: US. Government Printing Office: 211-231.

188

Grogger, Jeffrey (1991). “Certainty vs. Severity of punishment.” Economic
Inquiry. Vol. 29(2): 297 -313.

Harwood, H.J.; Napolitano, D.M.; Kristiansen, P.L.; and Collins, J.J. (1984).
Economic Costs to Society of Alcohol and Drug Abuse and Mental
Illness. Research Triangle Park, NC: Research Triangle Institute.

Hubbard, R.L., and Marsden, ME. (1986). "Relapse to use of heroin,
cocaine, and other drugs in the first year after treatment. " In:
Relapse and Recovery in Drug Abuse, edited by Tims, F .M.;
Leukefeld, C.G. National Institute on Drug Abuse Research
Monograph 72. Washington, DC: US. Government Printing Office:
157-166.

Kahn, HA. and Sempos, CT (1989). Statistical Methods in Epidemology.
New York: Oxford University Press.

Khattree, Ravindra and Naik, Dayanand N. (1995). Applied Mulitvariant
Statistics with SAS Software. Cary, NC: SAS Institute Inc: 70 - 91.

Klienman, P.A., Woody, G.E,; Todd, T.C.; Millman, R.B.; Kang, S,; Kempt.;
Lipton, D6. (1990). "Crack and cocaine abusers in outpatient
psychotherapy." In: Psychotherapy and Counseling in the Treatment
of Drug Abuse, edited by, Onken, L.S.; Blaine, J.D. National Institute
on Drug Abuse Research Monograph 104. Washington, DC: US.
Government Printing Office: 24-38.

Lipton, D.S.; Falkin, GP; and Wexler, H.K. (1992). "Correctional drug
abuse treatment in the United States: An overview." In: Drug Abuse
Treatment in Prisons and Jails, edited by Leukefeld, C.G.; Tims, F .M.
National Institute on Drug Abuse Research Monograph 118.
Washington, DC: US Government Printing Office: 8-30.

Lipton, Douglas 8., Gregory P. Falkin, and Harry K. Wexler. (1992).
"Correctional Drug Abuse Treatment in the United States: An
Overview." In: Drug Ause Treatment in Prisons and Jails, edited by
Leukefeld, C.G.; Tims, F .M. National Institute on Drug Abuse
Research Monograph 118. Washington, DC: US Government
Printing Office: 8-30.

Lurigio, A.J. (1987a). "Evaluating intensive probation supervision: The
Cook County experience." Perspectives 11: 17-19.

189

 

Lurigio, A.J. (1987b). "The perceptions and attitudes of judges and
attorneys toward intensive probation." Federal Probation 51:16-24.

Lurigio, A.J. and Davis, RC. (1990). “Does a threatening letter increase
compliance with restitution orders?: A field experiment.” Crime and
Delinquency 36 (4): 537-548.

Lurigio, A.J., and Petersilia, J. (1992). " The emergence of intensive
probation supervision programs in the United States." In: Smart
Sentencing: The Emergence of Intermediate Sanctions, edited by,
J.M. Byme, J.M.; Lurigio, A.J.; Petersilia, J. New bury Park CA:
Sage.

McLellan, AT, and Alterman, A1. (1991). "Patient treatment matching: A
conceptual and methodological review with suggestions for future
research." In: Improving Drug Abuse Treatment. National Institute on
Drug Abuse, edited by Pickens, R.W.; Leukefekd, C.G.; Schuster,
C.R. National Institute on Drug Abuse Research Monograph 106.
Washington, DC: US. Government Printing Office: 114-135.

Murray, Donald W.,Jr. (1992). "Drug abuse treatment in the federal bureau
of prisons: Initiatives for the 1990's." In: Drug Abuse Treatment in
Prisons and Jails, edited by Leukefekd, C.G.; Tims, F .M. National
Institute on Drug Abuse Research Monograph 118. Washington, DC:
US. Government Printing Office: 62-83.

National Institute on Alcohol Abuse and Alcoholism. (1990). "7th Special
Report to US. Congress on Alcohol and Health." Washington, DC:
US. Government Printing Office.

Pressman, J.L., and Wildavsky, A.(1984). Implementation: Third Edition.
Los Angeles CA: University of California Press.

Ridgely, MS, and Willenbring, ML. (1992). "Applications of case
management to drug abuse treatment: Overview of models and
research Issues." In: Progress and Issues in Case Management,
edited by Ashery, R.S. National Institute on Drug Abuse Research
Monograph 127. Washington, DC. US. Government Printing Office.

Ryan, James E. (1997). “Who gets revoked? A comparison of intensive
supervision successes and failures in Vermont.” Crime and
Delinquency 43(1): 104-1 18.

190

SAS Institute Inc. (1994). SAS/STAT Users Guide. Version 6 14th, Vol.2:
Cary, NC: SAS Institute INC:1187 - 1192.

SAS Institute Inc. (1995a). Logistic Regression Examples Using the SAS
System. Version 6, 1st edition: Cary, NC: SAS Institute INC:19 -157.

SAS Institute Inc. (1995b). SAS Procedures Guide. Version 6, 3rd edition:
Cary, NC: SAS Institute INC:339 - 430 and 621.

Sims, Barbara; Jones, Mark (1990). “Predicting success or failure on
probation: Factors associated with felony probation outcomes.”
Crime and Delinquency 43(1):314-327.

Swartz, J.A., Lurigio, A.J., and Scott, AS. (1996). "The impact of IMPACT:
An assessment of the effectiveness of a jail-based treatment
program." Crime and Delinquency 42(4):553—572.

U.S. Cong. House. Committee on Narcotics and Abuse and Control.
(Oct. 1991). Treatment and Rehabilitative Services for Substance
Abusing Criminal Offenders. 102nd Cong., 1st sess. HR. 730.
Washington, DC: US. Government Printing Office.

US. Cong. House. Committee on Narcotics and Abuse and Control.(1980).
United States Bureau of Prisons Staff Study: Institutional Drug Abuse
Treatment Programs and Utilization of Prescription Drugs at Five
Institutions. 96th Cong., 2nd sess. HR. 730. Washington, DC: US.
Government Printing Office.

US. Cong. House. Committee on Narcotics and Abuse and Control.
(1991). Hearing on: Treatment and Rehabilitative Sen/ices for
Substance Abusing Criminal Offenders. 102nd Cong., 1st sess. HR.
730. Washington, DC: US. Government Printing Office.

von Hirsch, Andrew. (1990). “The ethics of community-based sanctions.”
Crime and Delinquency 36(1): 163-173.

Wheeler John R.C. , Huda Fadel, and Thomas A. D'Aunno. (1992).

"Ownership and performance of outpatient substance abuse
treatment centers." American Journal of Public Health 82(5).

191

 

 

 

 

IIIIIIIIIIIIIIIIIIIIIIIII

~ IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII