lllllllllllllllll

 

 

 

 

 

 

1

This is to certify that the

thesis entitled

AN ANALYSIS OF RISK FACTORS
ASSOCIATED WITH PROBATIONER SUCCESS
IN
NEW JERSEY'S INTENSIVE SUPERVISION PROGRAM

presented by

MICHAEL T . COTTS

has been accepted towards fulfillment
of the requirements for

MASTER OF SCIENCE degree in WUSTICE

Date MAY 3, 1995

 

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

LIBRARY

Michigan State
University

 

 

 

PLACE ll RETURN BOX to remove thie checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU ie An Affirmative Action/Equel Opportunity inetitution
mane-9.1

 

AN ANALYSIS OF RISK FACTORS
ASSOCIATED WITH PROBATIONER SUCCESS
IN
NEW JERSEY'S INTENSIVE SUPERVISION PROGRAM

By

Michael T. Cotts

A THESIS

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

MASTER OF SCIENCE

Department of Criminal Justice

I996

ABSTRACT

AN ANALYSIS OF RISK FACTORS
ASSOCIATED WITH PROBATIONER SUCCESS
IN
NEW JERSEY'S INTENSIVE SUPERVISION PROGRAM

BY
Michael T. Cotts

The purpose of this study was to discern which offender risk
factors were most meaningful in the prediction of probationer
success (or failure) in an intensive supervision probation (ISP)
program. Secondary analyses were performed on a data set from
New Jersey‘s Intensive Supervision program, 1983-1986. Both
bivariate and multivariate techniques were utilized using SPSS. Ten
risk factors were drawn from four selected risk assessment models.
Using contingency table analysis and logistic regression analysis, the
findings suggest that employment status, number of prior
convictions, number of prior probation or parole revocations, as well
as race and sex were meaningful predictors of ISP outcomes. The
final logistic regression model correctly predicted 73 percent of ISP

probation outcomes.

ACKNOWLEDGMENTS

I wish to express my gratitude to my thesis committee and its
Chair. Dr. Charles Corley for their patience and persistence under
the constraints of time. For their technical assistance and expertise, I
wish to thank Mr. Brian R. Johnson and Dr. Randall Fotiu. Finally, for

her patient support, I give many thanks to my wife Carole.

TABLE OF CONTENTS

Chapter I
INTRODUCTION .................................................................................... I
Importance of Risk Assessment ............................................... I
Intensive Supervision Probation (ISP) ....................................... 2
Purpose of the Study ................................................................. 4
Potential impact for Probation Services ................................. 4
Chapter II
LITERATURE REVIEW ............................................................................. 6
Intermediate Sanctions ............................................................ 6
Intensive Supervision Overview ............................................... 8
Program Characteristics ............................................... IO
Fiscal Considerations............................................. ....... I2
Risk Assessment Overview ...................................................... I 4
The Wisconsin Model ..................................................... I 5
The Oregon Model ........................................................ I 7
The Colorado Model ..................................................... I 8
The New Jersey Model .................................................. I 9
Chapter III
METHODOLOGY ................................................................................ 22
Problem Statement ................................................................ 22
Description of Data Set .......................................................... 22
Selection of Variables ............................................................. 23
Contingency Table Analysis ................................................... 24
Regression Analysis (Logistic Regression) ............................. 24
Chapter IV
FINDINGS ........................................................................................... 26
Contingency Table Analysis ......................................... 26
Multivariate Analysis ...................................................... 34
Classification Table ........................................................ 38

Chapter V

SUMMARY AND CONCLUSIONS ...................................................... 39
Limitations of Risk Assessment ................................................ 39
Summary .................................................................................. 39
Conclusions .............................................................................. 41

APPENDIX A ....................................................................................... 43

APPENDIX B ....................................................................................... 47

BIBLIOGRAPHY .................................................................................. 48

Chapter I

INTRODUCTION

Importance of Risk Am

The subjective assessment of risk and the prediction of future
behavior have always been an integral function of our criminal
justice system. The ways in which individuals either suspected or
convicted of criminal activity are approached, processed, and
adjudicated are often based upon this assessment. Perhaps the
most important juncture for accuracy in risk prediction is at the
decision to release a convicted offender back into the community.
Classical madels have been developed to address the risk of future
criminal behavior, and the prediction of success or failure in the
community under various probation and pretrial release programs.
However, the assessment of risk of future criminal behavior based
upon past and current circumstances has always been a somewhat
elusive endeavor, as science can address and quantify only a finite
number of risk factors in an infinite realm of human behavior
patterns and settings.

While it is clear that the accurate prediction of potential
criminal behavior is a paramount goal for community-based
corrections as a social institution, it is also true that this goal carries
different weights at different levels of the criminal justice system. Risk
assessment will inevitably influence the techniques utilized by law
enforcement for apprehension and arrest. The assessment of risk
factors for bail and bond consideration issues may differ from

sentencing decisions and from the various classification schemes

utilized for those individuals who are placed under supervision in the
community or those sentenced to jail or prison.

Accuracy in predicting the risk of future criminal behavior at
each level and also in predicting the potential for positive change
dictates to a large extent how an offender will be 'processed' in the
system, and which sanctions will be imposed, and ultimately may
influence the degree of success an individual might attain under
such sanctions. The importance of risk assessment can hardly be
understated when considering the release of a traditionally prison-
bound offender back into the community. Inaccuracy could result
in increased expense, as well as increased incidence of recidivism.
Models of assessment have been developed and revised, and will
be further explored in this thesis. Their merits can only be as good as

their predictive successes.

Intensive Supervision Probation IISPI

Intermediate sanctions are those criminal sanctions that lie
between prison confinement and traditional probation. They have
been introduced and refined in recent years in response to
America's prison overcrowding dilemma. lnterrnediate sanctions
aim to reduce the enormous number of institutional commitments.
These sanctions offer to satisfy public desire for the criminal justice
system to act as a deterrent to criminal behavior. Several efforts,
such as Michigan's Community Corrections Act and other proposals
aimed at reducing prison populations nationwide, have left the
criminal justice system with the task of identifying the most

meaningful risk factors for the acceptance of an offender into the

community. Success and community acceptance of intermediate
sanctions as viable alternatives to incarceration depend heavily
upon the accurate assessment of risk. By study. application, and
change, research efforts assist in identifying those factors which will
best predict offender success under available intermediate
sanctions.

Perhaps the most popular and widely utilized intermediate
sanction in recent years has been Intensive Supervision Probation
(ISP). The balance between just-desserts punishment and cost-
effective prison-diversion have placed ISP programs in the spotlight
as an intermediate sanction. ISP programs have attempted to
bridge the gap between retribution, punishment, restitution, and
concerns over public safety on one side, and the over-utilization of
scarce prison space on the other. Though specific program
guidelines and requirements vary from one jurisdiction to another,
the basic premise behind ISP programs is to target those individuals
who would have been incarcerated under current sentencing
guidelines and/or practices, and provide for them an enhanced
supervision strategy in the community. As the name infers, the
programs are far more intensive in terms of probation supervision,
program requirements and personal responsibilities on the part of
the probationer than are required under more traditional probation
sentences. These 'high-risk' probationers may qualify for ISP either

before or after being sentenced to jail or prison.

Purpose of the Study

This study will focus on a specific area of community-based
corrections and Its population of offenders. Through a review of the
literature and a statistical analysis of selected risk variables chosen
from New Jersey's Intensive Supervision Program (ISP) an attempt will
be made to discern which risk factors represent the best predictors
of ISP program success. Success is defined as an unconditional
release from ISP supervision having met all program requirements
without violation. Failure is defined as a return to prison for either ISP
rule violation or a conviction for a new offense. Social, legal,
economic, psychological, vocational, and criminal history related
variables will be examined as they relate to the success or failure in
the New Jersey ISP program. As Latessa (I986) has pointed out:
“Indeed one of the problems facing ISP is the dilemma of
accurately selecting offenders appropriate for higher levels of
supervision in the community." Thus the two-fold purpose of
identifying predictive risk factors is first to refine eligibility and ISP
selection criterion: and secondly to develop more effective
supervision strategies for the probation officer in order to reduce the

incidence of offender recidivism.

Potenticilmpgct for Probation Services

The potential impact for the evaluation of risk assessment lies
in the decision-making processes at each juncture of ISP service
delivery. These junctures include not only initial acceptance into an

ISP program, but also involve decisions for the appropriate level,

type, and frequency of supervision, acceptance into various
rehabilitative services, and appropriate placements for community
service work. The notion of releasing potentially dangerous
probationers into our neighborhoods without adequate supervision
or community control has been met with concern from the public.
Likewise, if ISP is viewed simply as a net-widening sanction, then its
efficacy as a true diversion program may be diminished in the
criminal justice community. Thus, assessing risk and applying
sanctions based upon specific risk factors or variables has the
potential to re-shape public opinion, affect the acceptance of
community-based corrections, restructure sentencing guidelines and
practices, and ultimately reduce prison populations without posing
an elevated risk to the safety and security of our streets. It key
variables which play a statistically significant role in the success rate
of ISP probationers can be pinpointed, then the efforts to focus on
these elements to develop an enhanced supervision strategy may

be of significant benefit.

Chapter II
LITERATURE REVIEW

Intermediate Spnctions
The term "intermediate sanctions" refers to the range of post-
adjudication sanctions being developed to fill the gap between
traditional probation and traditional jail or prison sentences. The
following excerpt from the opening remarks by Dick Thomburgh,
then Attorney General for the United States. to the I990 National
Drug Conference, summarizes a philosophical rationale for
intermediate sanctions:
"...lf we recognize the gradations in the seriousness of criminal
behavior, then we should have gradations in sanctions as well.
That's why we need a portfolio of intermediate punishments
that are available-independent of whether our correctional
facilities are full or empty, or whether our correctional budgets
are lush or lean, or whether our offender populations are
increasing or declining." (US. Department of Justice, I990).
This suggested approach to the development of an array of
available sanctions is not the historical rationale for their creation.
The prevailing thrust behind these alternatives to incarceration has
been this nation's prison overcrowding dilemma.
Some of the points made by Michael N. Castle in writing for
the National Institute of Justice illustrate and illuminate the problem:
. Nationwide, about I in 50 persons is under the control of

correctional authorities.

. In the last decade, national per capita expenditures grew 2]

percent, but corrections expenditures grew 62 percent.

. The nation's prison population essentially doubled during the

1980's to more than 600,000 people.

. The growth of America's prison population is over ten times

that of the general population.

Accomplishing change and overcoming public resistance to
alternatives to incarcerating a felony population means putting an
end to the old-fashioned and inaccurate concept that criminal
justice means prisons and only prisons (Castle, I99I ).

Clearly, our nationwide problem of prison crowding
underscores the need for more intermediate sanctions, including
more effective probation supervision (Clear and O'Leary, I984). By
expanding sentencing options, intermediate sanctions enable the
criminal justice system to tailor punishment more closely to the
nature of the crime and the offender. An appropriate range of
punishments makes it possible to hold offenders appropriately
accountable for their actions. (Gowdy, NW). The range of
intermediate sanctions includes, but is not limited to, the use of
shock incarceration (boot camp), frequent drug testing, electronic
monitoring, intensive supervision probation, house arrest and home
confinement with or without electronic monitoring, day reporting
centers, expanded use of fines and restitution, community service,
and mandatory driver’s license suspension. Offender selection for
any one of these sanctions is based upon that individual's risk of

recidivism or potential for success in the community.

Intensive Supervision Overview

In the last twenty-five years. the role of probation in the field of
corrections has changed. It has been challenged to become the
core correctional process for most offenders (Clear and Shapiro,
I986). It is no secret that across the country, probation has a serious
image problem. Thomas Callahan, Director of Probation in New
York State. puts the problem in the vernacular: “People think that
probation is soft on crime." (Gettinger, I983).

There is also no question that probation departments are
dealing with a population that is involved in more predatory,
destructive, and dangerous criminal activity than at any other time
in probation's history (Cochran, I988). In order to maintain, and in
some cases restore its credibility as a pervasive social institution,
probation must amount to more than a suspended sentence, with
the probation officer reading monthly postcards from hundreds of
clients. Gettinger (I983) has asserted that making probation tough
on crime should be the primary goal of Intensive Supervision
programs, and that intensive supervision is what probation is
supposed to be. ISP programs are proliferating at a tremendous
pace in the wake of court-mandated solutions to prison crowding,
Increasing numbers of felony offenders sentenced to probation, and
increased public demand that probation agencies be held
accountable for these offenders.

Results of early evaluations seem to indicate the appeal of ISP.

Budget-conscious policy-makers. offenders who would otherwise be

incarcerated, the public concerned for its safety, courts in search of
viable sentencing alternatives, and corrections officials for whom ISP
offers an increased measure of credibility, all find in ISP a potential
solution to the concern for criminal justice accountability.

Most ISP programs are premised on the assumption that
certain high-risk/high-needs offenders can be handled more
effectively through an enhanced community supervision strategy.
(Cochran, Corbett, and Byme, I986). The overall purpose of ISP
programs is to protect the community in a cost-effective manner by
providing supervision via surveillance, and appropriate services for
offenders who, without such high-accountability direction, may be
incarcerated by a state's department of corrections.

Though many ISP programs differ in design, implementation.
and administration, most share many common characteristics in
pursuit of a more effective correctional sanction than regular
probation has traditionally provided.

The following discussion addresses the most common elements
of most ISP programs. It is certainly not an exhaustive or exclusive
listing of all features of all programs, but does address those aspects
which lay the groundwork for the majority of existing programs.
Unfortunately, "Intensive Supervision" is a “catch-all" phrase which
includes a wide range of pragrams at distinct decision points in the
criminal justice system. It has been used alternately to describe
programs which function as I) a front-end alternative to
incarceration (both in the form of a discretionary sentencing
decision controlled by a judge and as an established presumptive

term for a particular offense): 2) as a form of probation case

IO

management, once offenders are placed on regular probation
caseloads; and 3) as a “back-door" early release mechanism from
prison or jail (Byrne, I986). The "front door" approach of
implementing ISP conditions on those individuals either already
placed on probation, or those eligible for regular probation, has
come under fire for simply widening the net of social control. It has
been criticized as simply implementing yet another social control
mechanism on those who are already in the community. It is
contended that too many alternative programs have served merely
as add-ans to regular probation sentences (Austin and Krisberg,
I982). If ISP is to be viewed as a true diversion program, its offender
makeup should be of those individuals either sentenced or
committed to jail or prison. The problem, then, on'which subsequent
discussions in this thesis will focus, is targeting those individuals who
are truly prison bound. Clear and Shapiro (I986) suggest that once
a “true prison" group is eliminated because of crime seriousness and
risk, the incarceration bound offenders left come from a group of
offenders whose characteristics suggest the probability of
incarceration is only about 50 percent (Clear et al., I986). Thus, only
those “back door" programs which serve to release incarcerated
offenders or intercept commitment to jail or prison can escape the

criticism of net-widening considerations.

Program Characteristics
Intensive supervision as a concept is fairly straightforward. One
might expect that a probation officer would have a much higher

degree of control over an individual in the community through

II

heightened surveillance and frequent interactive contacts. Towards
the ends of reducing and eliminating criminal behavior on the part
of an individual convicted of illegal activity, the monitoring of the
present behavior of each individual is intensified. Most programs
require a minimum of five weekly face-to-face contacts between
the ISP officer and probationer, with strict enforcement of an
evening curfew. Some program designs incorporate a probation
officer to coordinate services and report to the court, and a para-
police surveillance officer to work in the field, on the street, making
contacts and enforcing curfews. Due to the intensity and frequency
of officer/offender contacts and interactions, caseload sizes vary
from as few as ten, to a maximum of fifty probationers, with twenty-
five being the target caseload for many programs. Most involve
frequent random drug and/or alcohol testing in the field to
determine whether an individual Is using or abusing controlled
substances. If substance abuse is an issue, then professional and
consistent treatment regimens are required for the probationer.
including up to daily Alcoholics/Narcotics Anonymous (AA/NA)
support group attendance. Professional intervention, counseling,
therapy, group therapy, educational or didactic sessions for anger
control, sex offender issues, domestic violence, and gambling
problems may all be required for the probationer. One may note
that these things have traditionally been included as part of regular
probation requirements. The difference is the Intensified degree of
accountability on the part of the probationer to the supervising
officer and ultimately to the court. By design, the deterrent effect of

having an officer so closely monitoring one's daily activities and

l2

holding one so strictly accountable for all behaviors should have its
most profound success due to the nature of the intensity. In terms of
sanctions and penalties, most ISP programs require heavy
supervision fees along with fines and costs per statute for the offense
committed. Often, monetary restitution to victims is required. Hence.
an offender is required to seek and maintain gainful employment
and/or complete job skills training, and frequently must complete a
minimum of up to hundreds of hours of community service work as

well.

Fiscal Considerations

From a fiscal standpoint, intermediate sanctions including ISP
are quite attractive to state and local budget reformers and policy
makers. Most are extraordinarily cost-effective when compared to
per capita prison commitment costs. Some estimates cite
approximately $16,000 per year to house a prisoner in a minimum
security camp or prison and $5,000 to provide intensive supervision in
the community, including $3,000 for the purchase of support services
such as substance abuse treatment. It is estimated that if the State
of Michigan could intensively supervise the 1,200 parolees and 2,400
probationers in the community who are currently going to prison, the
state could avoid building three additional prisons at a capital
outlay cost of $l25 million and an annual operating cost of $48
million. This compares with the cost of $I8 million for ISP for those
3.600 offenders (Michigan Criminal Justice Roundtable, I995).
Likewise, ISP programs in Oregon, Colorado, and New Jersey have

all shown very attractive cost benefit analyses. Evaluators of

I3

Georgia's ISP program have cited an average cost of $695 per
offender during their program period, compared to $25,215 per
offender had the offender been incarcerated (Latessa, I986).
Pearson (1990) in his evaluation of New Jersey's ISP program
estimated costs at $7,000 per offender year, while prison costs were
roughly $17,000 per offender year during 1986. Hence, with current
cost of imprisonment running between $20,000 and $30,000 per
offender per year, and the costs for new facilities running to $100,000
per bed, it is not surprising that many criminal justice professionals
are both optimistic and enthusiastic about the possibilities of ISP.

The Question of who is eligible for an alternative sanction
when faced with a possible sentence to confinement on a criminal
conviction is a problematic issue. Offender eligibility and specific
release criterion are key issues which must be explored in depth if ISP
is going to be effective in its overall goals of being cost-effective,
reducing prison populations, and satisfying the growing concern for
public safety. Pinpointing those risk conditions, variables, or factors
which contribute to ISP probationer success or failure is an essential
function of sentencing alternatives research. Target populations are
quite difficult to accurately and consistently define. Those most
frequently accepted into ISP are those felons who have committed
property crimes, theft, fraud, and/or drug offenses. Almost universally
excluded are cases of homicide, rape, and violent or aggravated
crimes. It is asserted that the primary aim of classification systems
and risk assessment instruments is to enhance supervision

effectiveness in the community (Smith. Rhine. and Jackson, 1989).

14

Risk Assessment Overview

The previous discussions on the philosophical, fiscal, and social
rationales for ISP programs as well as the general overview of basic
program structure has given rise to what must clearly be one of the
most critical aspects of ISP and all community-based corrections:
that is the appropriate selection of an offender population for such
programs. If the Criminal Justice system, as part of a larger society
deems it necessary to divert a traditionally prison bound offender
into a community, then the social as well as the specific research
question emerges: “Which individual risk factors are to be given
most weight, not only for an initial acceptance into an ISP program,
but also to focus on who is likely to be successful under intensive
supervision?" Several states have developed models for risk and
needs assessment for screening offenders for appropriate services,
sanctions, and levels of community supervision. But while various
labels are used to differentiate between levels of supervision, most
are essentially restatements of the traditional
Maximum/Medium/Minimum supervision classes (Nelson, Ohmart,
and Harlow, 1978).

Risk assessment and the scientific prediction of success in the
community can be divided into two broad categories: Clinical
methods and Statistical methods. The clinical approach considers
social, personality, and situational variables and the interactions of
these variables as the most essential elements of prediction. A high
degree of expertise is required to interpret such highly subjective
and personal elements. Simon (1971) indicated that reliability was a

major problem of clinical prediction due to variation, the clinician

l5

day by day, or clinician to clinician. She also indicated that the use
of mechanical data collection could be regulated by a specific set
of rules.

By contrast, statistical methods utilize mathematical formulas
to produce probability estimates of future behavior. Information
about an offender is compared with similar information on a large
sample of individuals whose behavior in similar situations is known
(Klofas, Stojkovic, and Kalinich, 1990). Statistical methods are
generally regarded as the more reliable and accurate for the
prediction of future criminal behavior. Sophisticated multivariate
models are most often employed to account for similarities and
differences in selected variables.

Four specific models of risk and needs assessment will be
examined in the following discussion. They are simplified overviews
of the assessment and/or classification models utilized by the
respective agencies in A) Wisconsin, 8) Oregon, C) Colorado. and
D) New Jersey.

The purpose of integrating the research efforts of each of these
projects is to compare the specific risk factors and variables which

were deemed most important in each model.

The WIscgnsin Mpciej

One of the most widely recognized and utilized risk
assessment tools is the Wisconsin risk assessment paradigm. In the fall
of 1977, an elaborate research project at a cost of more than one
million dollars was undertaken in Wisconsin. The Wisconsin

department of Corrections developed a classification system to

l6

assess a probationer's propensity for further criminal conduct and
assign them to different levels of supervision. The project was
considered the most methodologically rigorous caseload study
conducted in years, and the assessment of risk and needs had a
significant influence on respective probationer outcomes. The
Wisconsin system, also known as the NIC system, was also effective
in predicting success or failure in completing probation terms—low
risk cases were revoked at a much lower rate than the high risk ones
(Zhang and DeLaPaz, 1990). Classification or risk criterion in the
Wisconsin Model are as follows: I) Address changes in the past 12
months (the fewer changes of address reflect more stability in the
community): 2) Percentage of time employed in the past 12 months
(the more time employed, the more stability and responsibility
shown); 3) Alcohol usage problems: 4) Other drug usage problems
(drug and alcohol involvement may warrant professional
intervention and treatment to establish a chemical-free lifestyle): 5)
Attitude (an individual motivated toward success is expected to fare
better than one who is generally unmotlvated, angry, or displays a
poor attitude); 6) Age at first conviction (The lower the age, the
higher the risk, based upon the potential length of time in criminal
involvement); 7) Prior periods of probation or parole supervision: 8)
Number of prior probation or parole revocations (the greater the
number of unsuccessful experiences under community supervision,
the greater the risk of program failure): and 9) Number of prior felony
convictions (the greater the number, the higher the risk of failure).
Each of these variables is given a score based upon the

offender's criminal and social history. A raw score. a single number.

I7

is calculated from the individual's scores on each variable. That
number, in turn, gives the assessor and/or Judge a fair idea of how
much risk the offender poses to the community once released on

probation.

The Orpgon Maggi

Beginning in 1984, the Edna McConnell Clark Foundation
funded the Probation Development Project (PDP) in Multnomah
County, Oregon. Its two main objectives were first to develop a
probation unit which was capable of providing enhanced
supervision of offenders, and second to reform sentencing practices
to reduce the overall number of commitments to jail or prison. A
series of research tasks were undertaken to produce a plan for
identifying offenders for supervision. Researchers utilized a random
sample of felony offenders sentenced in Multnomah County within a
specified time frame. All variables in the screening model which
was developed resulted from multiple regression analysis of the
probation sample based on prediction of "failure". A total of 28
percent of the cases in the sample were failures. Multiple regression
analysis was used because research suggests that there is little
difference between this technique and other, more elaborate
methods of building prediction models (Gottfredson and
Gottfredson, 1984). A straightforward regression analysis of the
subsample produced the following variables. The criterion variable
was a success/failure dichotomy: ”Substance use: 2)Juvenile
convictions: 3)Needs at closing (on offender's rehabilitative.

vocational, and educational needs are assessed; the higher the

I8

degree of need, the higher the risk of failure): 4) Harm to victim
(greater degrees of physical and/or emotional harm may indicate
greater risk of failure): 5) Prior probation or parole revocations: and 6)
Age at first conviction.

Researchers did go on to note that after investigating the issue
,of racial bias, they Ieamed that ethnicity was not significantly
correlated with this scale. Interestingly, offense type was dropped.
as in no instance did it play a significant role in a client's eventual
performance. Thus, they were dropped from the final models.
Once the significant risk variables were codified, the individual was
given a raw “risk score". Raw scores then formed the Y-axis against
the X-axis of instant crime seriousness (as developed by the Oregon
parole board) forming a matrix displaying the percentages of cases
sentenced to probation by risk and crime seriousness. If an offender
is drawn from a program-eligible pool based upon risk assessment, a
standard presentence report (PSI) is prepared by the staff of a
diagnostic unit. If regular probation is about to be recommended by
the PSI writer, the case is not considered eligible by the PDP unit. If
the standard PSI results in a recommendation to an incarcerative
term, the case is forwarded to the PDP unit for consideration. A
recommendation to the program is then drafted for the sentencing

judges' consideration.

The Colorado Model
Similarly. the Colorado ISP program utilizes a sentencing matrix.
The purpose of the matrix is to identify prison-bound felons who

remain eligible for probation under Colorado statute. The use of the

I9

matrix. which is administered by probation officers, increases the
uniformity of the officers' sentencing recommendations to the court.
Risk factors or variables which are given weight in the Colorado
matrix are similar to those outlined in the previous models and
include: 1)Juvenile adjudications: 2) Juvenile commitments
(indicating a crime severity to warrant some term of incarceration):
3) Prior felony convictions: 4) Prior violent felony convictions: 5) Adult
probation revocations: and 6) Prior parole revocations (the two
revocation status's are separated in this model, with the heavier risk
of failure given to parole revocations).

As in the previous model, these weights are calculated for the
individual's criminal and social history and a raw score is assigned.
Based upon these risk criterion, the referring officer presents
information on the case including the matrix recommendation, and
any other relevant facts. 0 final recommendation to the court is

made by a probation officer prior to sentencing.

The New Jersey Mocfl

Finally, in his elaborate evaluation of New Jersey's Intensive
Supervision Program, Pearson (I990) used a stepwise regression
analysis to assess offender risk to recidivate in his comparison of an
ISP group with an ordinary term of imprisonment control group. The
variables of significance in this study we‘re: I) Instant offense type
(ranking for risk purposes from violent felony to property offense
misdemeanor): 2)Prior felony charges: 3) Race (Hispanic and black
having a higher failure rate than white non-Hispanic); 4)

Employment/unemployment: 5) Educational attainment (the higher

20

level of attainment, the lower the risk of failure); and 6) Age at time
of instant offense (the lower the age, the higher the risk score).

It should be noted here that in contrast to the first three models,
Pearson's variables of significance were used to compare recidivism
and success rates between groups, and not to determine who may
be eligible for the ISP program. Actual offender selection is made
by a panel of judges who resentence low-risk felons who have
served a minimum of three to four months in prison on their original
sentence. An evaluation of the program's effectiveness, including
both ongoing monitoring and comparisons between ISP participants
and a sample of similar felony offenders who served their terms and
were paroled, shows that ISP has been successful in meeting its
objectives. ISP saved prison space, reduced per offender
correctional costs, produced a high-rate (93.3 percent) level of
employment among participants, and reduced recidivism rates by
over 10 percent.

These four models of risk assessment were chosen for their
unique and contrasting approaches to predicting success. The
Oregon Model's variables seem to be overall more socially oriented,
whereas the Colorado Model uses a "rap sheet" approach, looking
at criminal history almost exclusively as its basis of prediction of
future criminality and appropriateness for supervision in the
community. The Wisconsin and New Jersey Models contain
trappings of both social and criminal history styles for their respective
predictive utilities.

In terms of assessment of risk of future criminality, it would seem

appropriate that an eclectic approach be used. Personal, social,

2]

and clinical issues alone have been viewed as insufficient for
purposes of prediction. Likewise, a “black and white" criminal
record leaves many life and potential success issues unexplored. It is
the marriage of both social and legal circumstances and individual
variables which should give rise to the most comprehensive picture
available when looking at an offender’s overall propensity to
recidivate or to be successful in the community under intensive
supervision.

There is some debate as to whether risk assessment results
should be presented to judges as part of presentence information, or
be used as a case management tool after a decision has been
made as to the appropriateness of an ISP sanction. There is no
single standard which can be established. since the philosophy
(e.g., just desserts, risk control, rehabilitation) and the purpose
(sentencing alternative, case management, prison crowding release
valve, etc.) of ISP programs vary from state to state, and within some
states as well (Byme, I986). Perhaps a combination of uses, for
sentencing, case management, and even ISP release eligibility

might be an appropriate approach for ISP refinement.

Chapter III
METHODOLOGY

Problem Statement

Toward developing an enhanced supervision strategy for
offenders in the community to reduce prison populations without
posing an elevated crime risk to the public, offender risk factors must
be analyzed to determine which felony offenders are most
appropriate for release. Statistical analysis will discern which risk
factors are most salient in the prediction of success as defined by

unconditional graduation from New Jersey's ISP program.

Description of Data Set
The selected variables have been drawn from the study

conducted by Pearson (1990). A general discussion of the
characteristics of the ISP participants chosen for supervision will
follow, and a more detailed description of each selected variable is
found in Appendix A.

The most serious of the crimes for which the ISP participants
received their instant sentence to prison was typically the distribution
of illegal drugs (43 %) or burglary (23 %I, together comprising most of
the ISP caseload. About two-thirds of the ISP participants had at
least one felony conviction before their instant sentence conviction.
About half of them had two or more prior felony convictions.
Approximately 30 percent of them served at least one sentence of

incarceration of more than thirty days before their instant offense.

22

23

Ninety percent of those accepted into the ISP program were male.
About 58 percent were white, 32 percent were black, and nine
percent Hispanic. These percentages are roughly what we might
expect to find in minimum security prisons in the United States as a
whole (Pearson, 1987). About half of those accepted into ISP had
neither a high school diploma or GED. About 28 percent of them
were unemployed at the time of their sentencing for their instant
offense. According to presentence report data, 57 percent had a
drug problem, and 29 percent had an alcohol problem. In eight
percent of the cases, a member of their family was noted in the
presentence report as having a criminal record. These facts suggest
that the ISP caseload is reasonably typical of the less-serious, less

violence—prone prison inmates in the United States (Pearson. 1990).

Selection of Variables

Based upon the previous discussions on the rationales and the
general acceptance of the risk variables in the preceding models,
the independent variables including basic demographic
characteristics which will be analyzed for their predictive utility in this

study are listed below.

e SEX

e RACE

e OFFENDER’S LIVING SITUATION IN THE COMMUNITY
e EMPLOYMENT STATUS AT SENTENCE

e DRUG DEPENDENCE PROBLEM

e ALCOHOL DEPENDENCE PROBLEM

24

e EDUCATION COMPLETED

e NUMBER OF PRIOR CONVICTIONS

e NUMBER OF TIMES ON PROBATION/ PAROLE

e NUMBER OF PROBATION/ PAROLE REVOCATIONS

These variables represent offender factors drawn from the
literature which may lead to more effective supervision strategies
once they take their respective places in a scheme of predictive

accuracy and strength.

Contingency Tgple Anqlvsis

The Chi-squared evaluation of contingency tables is an
especially useful analytical device (Kachigan, 1986). Each
independent variable is crossed with the dependent variable in
order to assess the frequencies for ISP graduation and return to
prison within the parameters of the risk factor in question. Findings
are graphically plotted in 2 x 2 or 2 X k tables with the chi-squared
figure within the appropriate degree of freedom (dF) given. Phi
values as well as Spearman's Rho values are presented as indicators
of the strength of association between each independent variable

and its respective outcome.

R_eg§ssion Analysisllppgistic Regression

Regression analysis is a general method of analyzing much
behavioral research data. Its established utility in the assessment of
the relative degree to which predictor variables account for

variance in a criterion (dependent) variable makes regression

25

analysis a desirable method for this study. When using a
dichotomous dependent variable (successful/unsuccessful ISP) it is
preferred to use a non-linear Iogit regression model (Knapp, Kostas,
and Missiakoulis, 1982). The categorical and continuous predictive
variables as they take their place in the research question "To what
degree is variable X significant to successful ISP completion?" have
been collapsed and re-coded where possible for purposes of
analysis (refer to Appendix A).

Simply stated, the analysis of the available data will determine
which risk factors are the strongest predictors of success in New
Jersey's ISP program. The selection of the independent variables
from the available data is based upon the previous discussion on
the value of using both criminal history and personal/social histories.
The dependent variable in the equation will be
successful/unsuccessful completion of ISP probation as defined by

the ISP data values of either A) graduated, or B) returned to prison.

Chapter IV
FINDINGS/RESULTS

Confipgenpv Table Analysis
In the following discussion each independent variable is

displayed in a contingency table (2 X 2 or 2 X k). Raw numbers are
shown in each cell with their respective percentages In parentheses.
The dependent variable remains consistent throughout as
Graduated or Returned to prison. The apprOpriate chi-square values
and association measurement statistics are discussed for each of the
variables.

Table I shows the contingencies for the variable SEX. A larger
percentage of women admitted to the ISP program successfully
completed the program. 69 percent of the women were
discharged successfully from the program. 57 percent of the men
graduated successfully. The chi-square Value of 2.25 at the .13 level
of significance does not indicate a particularly significant difference
between the sexes, nor does the Phi value of .069 indicate a

meaningful level of association between SEX and success or failure.

Table I

Graduated or Rptumed to m‘son by SEX

 

 

Male Female
Returned 1 79 (43%) 15 (31)
Graduated 242 (57) 33 (69)

 

 

 

 

Chi-Square=2.255 DF=I Significance=.1331 4
Phi Value=.06935 Significance=.13314

26

27

In the assessment of RACE as it relates to the incidence of
graduation from ISP, the chi-square value of I9.22 with a .0000l level
of significance indicates a strong significance for the RACE variable.
The Non-white cell reflects black, Hispanic, and all other orientations.
Table 2 shows that 67 percent of white ISP participants graduated,
while 47 percent of all non-white races did likewise. A Phi value of
.20 however, indicates only a weak to moderate measure of

association between graduation and RACE.

Table 2

Graduated or Returned to prison by RACE

 

Non-White White
Returned 105 (53%) 89 (33)
Graduated 93 (47) I82 (67)

Chl-Square= 19.226 DF=1 Significance=.00001
Phi Value=.20247 Significance=.00001

 

 

 

 

 

 

Table 3 illustrates that in terms of an offender's living situation in
the community, the highest levels of success in ISP are for those
individuals living either with spouse and child or one or the other. 62
percent of those who live either alone or with a sibling were
successful. Those prObationers living with friends and those living
with parent and sibling each had roughly a 50 percent chance of
success in the program. The Chi-square value of 19.05 at the .00077
level reveals that an offender's living situation is indeed significant.
However, the Phi value at .20158 demonstrates a rather weak
association between each living situation in this categorical variable

and graduation from ISP.

28

Table 3

Graduated or Returned toprison by LIVING SITUATION IN COMMUNITY

 

 

Friends Parent/ Alonel Spouse or Spouse

Sibling Sibling Child and Child
Returned l6(52%) 114(49) 22(38) 14(29) 28 (27) |
Graduated 15(48) 116(51) 36 (62) 35 (71) 73 (73) |

 

 

 

 

 

 

Chi-Square=l9.057 DF=4 Significance=.00077
Phi Value=.20158 Significance=.00077

Data on an offender's employment status at the time of
sentence was gathered and is presented in Table 4. Nearly 70
percent of those with full-time employment at the time of their
sentence were successful in ISP. By contrast, a full 63 percent of
those individuals who were unemployed or unable to work were
returned to prison for failure in the program. These figures must be
viewed with caution, as they represent employment status at
sentence, and not necessarily status at the time of program failure.
This fact is noteworthy here, as unemployment represents the
variable with the highest probability for Returned to prison in the
multivariate analysis which will be discussed in the next section.
There is an extraordinary significance level at .00000 for this variable,
with the chi-square value at 40.5 with 2 degrees of freedom.
Spearman's Rho value is used as a measure of association for this
ordinal variable, and its value of .285 would indicate a slight to

moderate association between employment and successful

29

completion of the ISP program. The relationship at the .00077 level

does however show statistical significance.

Table 4

Graduated or Returned to 'son b EMPL YMENT AT AT ENTEN E

 

 

Not Part Time Full time
Employed Job Job
Student
Returned 88 (63%) 19 (40) 87 (31)
Graduated 51 (37) 29 (60) I95 (69)

 

 

 

 

 

Chi-Square=40.5l3 DF=2 Significance=.00000
Rho Value=.28490 Significance=.00000

Drug and alcohol dependence problems and their respective
rates of success and failure are outlined in Table 5 and Table 6. In
each table, an offender's likelihood of being returned to prison or
graduating successfully from ISP is fairly evenly split for those with
dependence problems. For drugs, a 55 percent success rate: for
alcohol, a 5] percent success rate. For those without drug habits, a
64 percent success rate is realized. For those not diagnosed alcohol
dependent, 62 percent were successful. Given these splits, it follows
that the respective Chi-square values are significant only at .035

and .029 levels, with very slight association values on the Phi statistic.

Table 5

Graduated or returned to prison by DRUG DEPENDENCE PROBLEM

 

 

Drug No
Problem Problem
Returned 124 (45%) 70 (36)
Graduated I49 (55) 126 (64)

 

 

 

 

Chi-square=4.432 DF=1 Significance=.03527
Phi Value=.0972l Significance=.03527

30

Table 6

Graduated or Returned to prison by ALCOHOL DEPENDENCE PROBLEM

 

 

Alcohol No
Problem Problem
Returned 72 (49%) 122 (38)
Graduated 76 (51) 199 (62)

 

 

 

 

Chi-square=4.730 DF=1 Significance=.02964
Phi Value=.10043 Significance=.02964

In terms of an offender's level of educational attainment,
Table 7 indicates a clear and significant correlation between
educational attainment and success in the ISP program. Those who
dropped out of High School had less than a 50 percent success rate,
as opposed to those individuals with higher levels of attainment such
as High School graduation or GED completion, which shows a 68
percent success rate, and the highest levels of attainment (some
college or college degree) reaching a 74 percent level of successful
completion of the program. The Spearman's Rho value
approaching .24 would indicate a slight to moderate degree of
association between levels of attainment and ISP graduation in this
categorical variable. The first category (other) is a result of the
collapsing of the other four categories and represents a small
number of odd combinations of partially completed educational

endeavors at various levels.

3I

 

 

 

Table 7
Graduated or Returned to prison by EDUCATION COMPLETED
Some
Other HS Dropout HS Grad/ College] BA.BS
GED VocTech PostGrad
Returned 3 (60%) 126 (52) 38 (32) 22 (26) 5 (26)
Graduated 2 (40) 115 (48) 81 (68) 63 (74) 14 (74)

 

 

 

 

 

 

 

Chl—Square=27.098 DF=4 Significance=.00002
Rho Value=.23702 Significance=.00000

As has been demonstrated in the risk assessment models
discussed earlier, one of the most salient risk evaluation factors is an
offender's criminal history as indicated by his or her number of prior
criminal convictions. Table 8 displays the raw numbers and their
corresponding percentages of those ISP participants who were
successful in relation to their criminal conviction records. An inverse
proportion is clearly identified in this continuous variable by a Rho
value of -.3231 6. Those individuals with no prior convictions enjoyed
the highest level of success at 76 percent, followed by those with a
single criminal conviction at 66 percent. ISP participants with two,
three, or four convictions, comprising roughly 30 percent of the ISP
population had nearly an equal split between success and failure
rates with 56, 47, and 47 percent successful, respectively. Those with
five prior convictions showed the poorest success level at only 18
percent. Table 8 shows that those with six or more convictions.
comprising nine percent of the total ISP population had an overall

success rate of just 34 percent.

32

Table 8

Graduated or Reiumed to prison by NUMBER OF PRIOR CONVICTIONS

 

 

O 1 2 3 4 5 6+
Returned 35 (24) 33 (34) 31 (44) 21 (53) 17 (53) I4 (82) 43 (66)
Graduated 111 (76) 65 (66) 4O (56) I9 (47) 15 (47) 3 (18) 22 (34)

 

 

 

 

 

 

 

 

 

Chi-Square=58.268 DF=16 Significance=.00000
Rho=.32316 Significance=.00000

Likewise, a nearly universal criminal history factor which is
shown to be nearly as significant as the number of prior convictions
is an individual's number of times he or she has been placed on
probation as a matter of sentence. The strongest inverse relationship
between the numbers of probation terms is shown by a Spearman's
Rho value of -.27158. The significance level is also meaningful at
.00015. Table 9 shows that those ISP participants who never served
any term of probation or parole had the highest overall success rate
at 73 percent, followed downward in success rates by those who
had served one, two, three, and four terms or probation or parole
with success rates at 57, 46, 42, and 31 percent, respectively. Those
with five prior terms graduated at a 44 percent rate, and those with
six or more prior terms of probation or parole which comprised only
five percent of the total ISP population graduated at a rate of just 35

percent.

33

Table 9

Graduated or returned to prison by NUMBER OF TIMES ON PROBATION] PAROLE

Returned

Graduated

 

 

O I 2 3 4 5 6+
53 (27) 59 (43) 32 (54) 21 (58) 9 (69) 5156) 15 I65)
140 L73) 78 (57) 27 (46) 15 (42) 4 (31) 4 (44) 7 (35)

 

 

 

 

 

 

 

 

Chi-Square=39.705 DF=I3 Significance=.00015

Rho Value=.27158 Significance=.00000

Table 10, which shows success and failure rates by the number

of prior probation or parole revocations, demonstrates a similar

inverse relationship with a comparable Rho value at -.21885 and

significance at .00015. 64 percent of those ISP participants who had

never had a probation or parole term revoked were successful in

ISP. Conversely, those with four or more revocations, comprising just

one percent of the ISP population, were returned to prison from the

program at a 100 percent failure rate.

Table 10

Graduatged or Relumed to prison by NUMBER OF PROBATION/PAROLE REVOCATIONS

 

 

0 1 2 3 4+
Returned 139 (36%) 37 (62) 7 (34) 5 (83) 6 (100)
Graduated 244 (64) 21 (38) 9 (56) 1 (17) 0

 

 

 

 

 

 

Rho Value=.2l885 Significance=.00000

Chi-Square=28.991 DF=7 Significance=.00015

 

 

34

Multivariate Analysis

To further understand the impact of the aforementioned
variables (sex, race, living situation, employment status, drug
problem, alcohol problem, education level, number of prior
convictions, number of times on probation or parole, and number of
probation or parole revocations) a multivariate statistical technique
was introduced. Logistic regression was employed to assess the
impact of each independent variable in the larger scheme of all of
these variables an ISP graduation. As indicated in Chapter 3, logistic
regression is well suited to many kinds of data frequently found in
sociological research. So many of the dependent variables of
interest are dichotomous in nature, while many of the independent
variables impacting on them are measured at other levels (Walsh,
1987). The variables in this study are set up in just this way. The
dependent variable, Graduated or Returned to prison, has a
dichotomous coding of I and 0 for their respective outcomes. The
coding schemes for the ten independent variables, which have
been enumerated in previous discussions, are shown in Appendix A.

The data displayed in Table II is the result of a stepwise
logistic regression analysis run in SPSS for Macintosh PC version 6.1.
The tabular results shown are displayed in rank order based upon
the weight of the Beta value of each variable. The reader will note
that for the categorical variables (employment status, educational
attainment, and living situation) the SPSS program, in building this

predictive odds model, selected one category from each of these

35

Table II
Stepwise Logistic Regression

 

 

Variables In the Equation

 

 

Variable 8 SE. Sig R
Unemployed/ U nable -.7256 .1835 .0001 -.1473
Live w/ Friends -.6785 .3882 .0804 -.0410
HS Dropout -.4I92 .3012 .1640 .0000
Live Alone/Sib -.3457 .1902 .0691 -.0456
Num Prior Revocations -.3761 .1749 .0315 -.0646
Num Prior Convictions -.1589 .0414 .0001 -.1424
HS Grad -.1010 .9733 .9173 .0000
Live w/ Parent/Sib -.0415 .2800 .8822 .0000
Some Coll/VocTech .2238 .3345 .5034 0000
Full Time Employment .2869 .2413 .2346 .0000
College/PostGrad .3950. .3548 .2655 .0000
Live w/ Spouse/Child .4867 .3181 .1260 0233
Race .7746 .2239 .0005 .1 260
Sex I .1 430 .4180 .0062 .0934
Constant .4292 .3462 .21 51

Variables Not In the Equation
Variable Score DF Sig R
Alcohol Dep Prob .1102 I .7399 .0000
Drug Dep Prob 1.8912 1 .1690 .0000
Times on Probation .1202 1 .7288 .0000

variables and incorporated this into the constant. Each independent
variable was run in the deviation contrast mode, pitting it against
each of the other variables in the equation. The program internally
selected those variables which were deemed significant in relation
to the others, and excluded those which were not (alcohol problem,

drug problem, and number of times on probation or parole). As is

36

the case with multiple regression, the contribution of individual
variables in logistic regression is difficult to determine. In the contrast
mode presented, the contribution of each variable depends on the
other variables in the model. This is a problem, particularly when
using a large number of independent variables which are highly
correlated.

In terms of the (coefficient) Beta value interpretation, the
higher Beta scores increase the log odds of success or Graduation
from the ISP program, coded 1. The given strength of the negative
Beta values increase the log odds of failure, or Returned to prison,
coded 0. Thus we can see that unemployment, living with friends,
High School dropout, living alone or with sibling, number of prior
probation or parole revocations, number of prior convictions, High
School graduation, and living with a parent or sibling, have, in
varying degrees, stronger propensities toward failure based upon
their coding. Indicators and their strengths for success or Graduation
from ISP were some college or vocational education, full time
employment, college degree, and living with spouse and child. In
terms of race and sex, positive Beta values demonstrate a higher
propensity for whites and females toward success (each coded 1,
with their respective counterparts coded 0). The reader will note the
disproportionate Beta value for the predicted success for the
variable SEX at the value 1 (female). This may be due in part to the
very low number of females in the program (N=48) and their higher
success rate than males (N=421).

However, this model as a whole should be interpreted with

caution, as the significance levels for the majority of variables in the

37

model lie at unacceptable levels. Those variables selected by the
SPSS logistic regression model as significant at the .05 level or better
were employment status (unemployed) number of prior probation or
parole revocations, number of prior convictions, sex, and race.

In order to assess consistency and magnitude of significance
of these five variables, they are compared and contrasted to their
respective counterparts in the bivariate analysis. Employment status
shows extraordinary significance at .00000. The category of interest
(unemployed) shows a 63% return to prison rate. This is consistent
with the strong negative Beta weight in the multivariate analysis. The
number of prior probation or parole revocations also proves to be
very significant In the bivariate analysis at .00000 with a strong
relationship between higher numbers of revocations (4+ revocations
were returned to prison at a 100% rate) - consistent with the negative
Beta in the regression table. Likewise, number of times on probation
or parole was also significant at the .00000 level with high numbers
of probation or parole experiences showing a failure rate of 65
percent, also consistent with the negative Beta score. The
dichotomous variable sex, however, did not attain adequate
significance in the bivariate analysis at .13314. This again may be
explained in part by the disproportionate number of males to
females and the inverse relationship between the percentage of
females in the ISP program and the higher success rates they
attained. Finally, race showed a significant level in the bivariate
analysis at .00001 with a success rate at 67 percent for value 1

(white, non-Hispanic).

38

Thus, with the exception of the variable sex, the independent
predictor variables which attained acceptable significance levels
in the multivariate predictive model were highly consistent with the

straightforward findings shown in the bivariate contingency tables.

Table 12

Classification Table for Graduatginetumed to prison

 

 

 

magi.
Observed Returned Graduated
Returned 1 15 75 60.53%
Graduated 51 223 81 .39%

 

 

 

Overall 72.84%

Table 12 shows that the logistic regression model with the
variables selected by the SPSS contrast function successfully
predicted 81.39% of ISP graduates and 60.53 of those returned to

prison, for an overall success rate of 72.84%.

Chapter V
SUMMARY AND CONCLUSIONS

imitptions of Risk Assessment

The assessment of offender risk for supervision in the
community must always be made with caution. The development
and continued refinement of risk assessment scales must be firmly
ground in empirical data and employed as a single tool in a larger
milieu of crime seriousness, community resources, and sound judicial
discretion. There is an inherent danger in attempting to predict
future behavior based upon past behavior and experience. Life
itself has such a fluid dynamic, ever-changing and evolving, with an
infinite number of minute-to-minute choices and challenges. To
confine an offender to his or her past to influence future behavior
may be a disservice to that individual and to the community as well.

In a balanced approach to ISP probation, administrators have
to deal with the unfortunate reality that some probationers will see
change as a burden instead of an opportunity. Risk reduction based
upon accurate risk assessment, assumes that each person has the
potential to learn, to change, and to become a law-abiding and
productive participant in the life of the community. This assumption
can be measured only in terms of the progress (or lack thereof) of

the individual once sanctions are in place.

Sum
The prison crowding dilemma in the United States has given

rise to an array of intermediate sanctions which lie between prison

39

4O

commitment and traditional probation supervision. Intensive
Supervision Probation (ISP) programs have proliferated and lie at the
forefront of such intermediate sanctions. Intensive supervision Is
intended to successfully supervise an offender in the community
through greatly heightened surveillance, enforcement of strict
curfews, frequent random drug and alcohol testing, coordination of
rehabilitative services, and requiring full time employment or
education and/or vocational training. ISP probation officers have
smaller caseloads, averaging 25 offenders per officer, making
possible the enforcement and oversight of such stringent
requirements, and ultimately placing greater accountability on the
offender for his or her actions in the community.

Clearly, one of the most critical issues in the implementation of
ISP or any intermediate sanction which seeks to divert a traditionally
prison-bound offender from the institution to the community is the
issue of offender selection. The question of who is eligible and who is
not eligible and the reasoning behind such determinations very
often holds the key to offender success or failure in the community.
There have been several assessment devices from the simple to the
very complex, which have been devised and revised for purposes of
screening offenders for risk of recidivism in the community. Variations
of such devices are used by booking agents for bond
considerations, probation officers during presentence interviews,
judges at sentencing, parole boards of parole release hearings, and
probation and parole officers in developing supervision strategies.
Many of the risk screening devices have several of the personal and

criminal history items in common.

41

Four such devices were selected from the risk assessment
literature for consideration in this thesis. Models utilized by the
respective correctional agencies in Wisconsin, Oregon, Colorado.
and New Jersey were described and discussed. From these popular
models, an eclectic approach was used to incorporate both
criminal history and personal and social risk factors to determine a
common list often risk factors to be used as variables in this study.

Data from New Jersey's Intensive Supervision Program was
obtained through the Inter-University Consortium for Political and
Social Research (ICPSR). The research problem to be applied to the
available data was to determine, through statistical analysis, which
of the selected variables were most meaningful in the prediction of
offender success or failure in the ISP program. Using both bivariate
and multivariate analyses, the variables which were significant at
acceptable levels for both methods were employment status,
number of prior convictions, number of prior probation or parole
revocations, sex, and race. The logistic regression model run through
SPSS had an overall correct prediction rate of nearly 73 percent,
indicating that ISP offender selection based upon models which
assess offender risk may account for much higher predicted
outcomes than offender selection based merely upon chance

alone.

Conclusions
The development of criminal sanctions which sufficiently
punish an offender and deter that offender from future criminal

activity, while simultaneously serving to rehabilitate that individual so

42

that he or she can enjoy a crime-free lifestyle will undoubtedly
continue to be a difficult endeavor. Faced with the current trend in
prison crowding, pitted against a burgeoning public sentiment to
get tough on crime, criminal justice agencies nationwide are
challenged to foster and improve alternatives to incarceration
which met these lofty demands.

Intensive Supervision programs in Georgia, Massachusetts,
Arizona, and several other local programs as well as the New Jersey
model have shown favorable resUIts (Qt’erms gfisaving prison space,
saving correctional dollars, reducing recidivism rates, compensating
victims, and rehabilitating offenders. By carefully screening offender
risk and eligibility for ISP programs, the positive aspects of offender
rehabilitation and accountability along with public demands
associated with community supervision may be more easily realized.
Once again, however, this will inevitably present a formidable
challenge to criminal justice professionals who wish to attain these
most favorable outcomes. By way of research-based planning and
development of ISP programs, these sanctions may eventually
evolve into effective and accepted means of dealing with a

growing felony population.

APPENDICES

APPENDIX A

Selected Variables and Frequencies

43

APPENDIX A
Selected Variables and Frequencies
A) DEPENDENT VARIABLE

GRADUATED OR RETURNED TO PRISON

Value Label Value Frequency Percent
Returned to Prison 0 194 41.4
Graduated I 275 58.6

 

 

8) INDEPENDENT VARIABLES

SEX (N=469)

Value Label Value Frequency Percent
Male 0 421 89.8
Female 1 48 10.2

 

RACE (N=469)
Value Label Value Frequency Percent

Non-white 0 1 98 42.2
White 271 57.8

d

44

OFFENDER'S LIVING SITUATION IN THE COMMUNITY (N=469)

 

 

 

Value Label Value Frequency Percent
Friends 0 31 6.6
Parent/ Sibling 1 230 49.0
Alone/Just Sibling 2 58 12.4
Just Spouse/Just Child 3 49 10.4
Spouse/Child 4 101 21 .5
EMPLOYMENT STATUS AT SENTENCE (N=469)

Value Label Value Frequency Percent
Unemployed/Unable 0 139 29.6
Part-lime Job/Student I 48 10.2
Full-Time Job 2 282 60.1
DRUG DEPENDENCE PROBLEM (N=469)

Value Label Value Frequency Percent
Yes 0 273 58.2

No 1 196 41.8
ALCOHOL DEPENDENCE PROBLEM (N=469)

Value Label Value Frequency Percent
Yes 0 148 31.6

No I 321 68.4

45

EDUCATION COMPLETED (N=469)

 

Value Label Value Frequency Percent
Other 0 5 1.1
HS Dropout 1 241 52.5
HS Grad/GED ' 2 119 25.4
Some College/HSBVocTech 3 85 18.1
BA/BS/Post Grad degree 4 19 4.1
NUMBER OF PRIOR CONVICTIONS (N=469)
Value Label Value Frequency Percent

0 146 31.1

1 98 20.9

2 71 15.1

3 40 8.5

4 32 6.8

5 17 3.6

6 22 4.7

7 6 1.3

8 9 1.9

9 10 2.1

10+ 18 3.8

46

NUMBER OF TIME ON PROBATION OR PAROLE (N=469)

Value Label

Value

d‘OCDVCfiUt-hOOM—‘O

O
+

Frequency
1 93

1 37

59

36

1 3

(AM—'Ut‘OO

Percent
41 .2
29.2
12.6

7.7

2.7

I .9

1 .9

1 .

 

NUMBER OF PROBATION/ PAROLE REVOCATIONS (N=469)

Value Label

Value

#00de

Frequency

383
58

1 6
6

6

Percent

81 .6
12.4
3.4
1 .3
I .3

APPENDIX 8

Assistance/ Disclaimer Statement

47

APPENDIX B

Assistgnce/Disclaimer Statement

The Data and tabulations utilized in this thesis were made available
by the inter-University Consortium for Political and Social Research.
The data for INTENSIVE SUPERVISION PROGRAM IN NEW JERSEY.
1983-1986 were originally collected and prepared by Frank Pearson.
Neither the collector of the original data nor the Consortium bears

any responsibility for the analyses or interpretations presented here.

BIBLIOGRAPHY

48

BIBLIOGRAPHY

Austin, Jim, and Krisberg, Barry (1982). The Unmet Promise of
Alternatives to lnacrceration. San Fransisco: Nptionpl Council
on Crime 93;! Delinguency: Mimeo.

Baird. S. Christopher (1983). Intensive Supervision in Probgtion: A
Survey of Current Programs. Madison. WI: National Council on
Crime and Delinquency.

Burkhart, W.R. (1986). Intensive Supervision Probation: An Agenda
for Research and Evaluation. Federal Probation, fl (2). 75-78.

Byme, J.M. (I986). The Control Controversy: A Preliminary
Examination of Intensive Supervision Programs in The United
States. Federal Probation, fl (2). 4-19.

 

Castle, Michael N. (1991). Altemgtive Sentencing: Selling It to the
Public. Washington, DC: National Institute of Justice
Research In Action

Clear, TR. and Shapiro. C. (1986). Identifying High-Risk
Probationers for Supervision in the Community: The Oregon
Model. Federal Probation 5_0_ (2), 42-49. ‘

Clear, Todd, and O'Leary, Vincent (1984) Controlling the Offender
in the Community, Lexington, Massachusetts: Lexington
Books. Pp.79.

Cochran, D. (1988). Corrections' Catch-22. Corrections Today.
October, 1988, 16-19.

Cochran. D., Corbett, RP, and Byme, J.M. (1986). Intensive
Supervision Probation in Massachusetts: A Case Study In
Change. Federal Probation, 59 (2). 32-41.

Fogg, V. (1988). Effective Alternative: Colorado's Intensive
Supervision. Corrections Today, 50 (1), 50-58.

49

Gettinger, S. (1983). Intensive Supervision: Can It Rehabilitate
Probation? Corrections Magazine. April, 1983. Pp. 6-17.

Gottfredson, S. and Gottfredson, BM. (1984). Sgeening for Risk.
Washington, DC: National Institute of Corrections.

Gowdy, Voncille (1991). Intermediate Sanctions. Washington DC:
National Institute of Justice Research in Brief.

Kachigan. S.K. (I986). SLJtistlpgl Anglysis. New York: Radius Press.

Kerlinger. F. (1986). Foundations of Behgvioral Resegrch, Fort Worth:
Holt, Reinhart and Winston, Inc.

Klofas, Stojkovic, and Kalinich (I990). Criminal Justice
Organgitions: Administrgtion andr Management. Pacific
Grove, CA: Brooks/Cole Publishing.

Knapp, M. Kostas, H. and Missiakoulis, S. (1982). Investigating labor
turnover and wastage using Logit Technique. Journal of
Occupational Psychology, 55. 129-138.

Latessa, E.J. (1986). The Cost Effectiveness of Intensive Supervision
Probation. Federg_l Probation, 50 (2). 70-81.

Lindner, C., and Kohler, R. (1992). Probation Officer Victimization:
An Emerging Concern. Jourpgl of Criminal Justice. 2_0. 53-62.

Lipchitz. J.W. (1986). Back to the Future: An Historical View of
Intensive Probation Supervision. Federal Propation, & (2), 78-
81.

Lurigio, A.J. (1987). Perceptions and Attitudes of Judges and
Attorneys Toward Intensive Probation Supervision. Federal
Probation, .5_1 (I). 16-24.

 

Neubauer. D.W. (1988). Americg's Courts and the Criminal Justi<_:_e_
System. Pacific Grove, Ca: Brooks/Cole Publishing.

 

50

Michigan's Criminal Justice/Substance Abuse Rountable (1995).
Time for a Change: A Strategic Framework IDrcmL.

Morgan, S. Philip and Teachman, JD. (1988). Logistic Regression:
Description, Examples. and Comparisons. Joumfiof Marriage
M the Fpmily. 50, 929-936.

Nelson, E.K., Ohmart, H., and Harlow, N. (1978). Promising Strategies
in Probation and Parole. Washington DC: National lntitute of
Law Enforcement and Criminal Justice.

Pearson, Frank S. INTENSIVE SUPERVISION PROGRAM IN NE_V1
JERSEY. 1983-1986 (Computer filel. New Brunswick, New
Jersey: Institute for Criminological Research, 1987. Ann Arbor,
MI: Inter-University Consortium for Political and Social
Research, 1990.

(1987). Final Report of Resegrch on New Jersey's Intensive
S_upervislon Program. Rutgers University Institute for
Criminological Research.

 

(1988). Evaluation of New Jersey's Intensive Supervision
Probation. Crime andDelinauency. 34 (4), 437-448.

 

Simon, F. H. (1971). Prediction Methods in Criminology. London: Her
Majesty's Stationary Office.

Smith, Wm., Rhine, E., and Jackson. R. (1989). Parole Practices in
the United States. Corrections Todgy. October 1989, Pp. 22-28.

SPSS User's Manual 6.1.1 (1994) Chicago. Ill: SPSS.

Tabackhnick, B.G. 8. Fidell, LS. (1989). Using Multivariate Stgtistics
(2nd ed.). New York: Harper Collins.

Walsh, Anthony (1987). Teaching Understanding and
Interpretation of Logit Regression. Teaching Sociology, 1_5_,
178-183.

51

Weniz, C. and Oldroyd, R.J. (1979). Rational Risk Assessment for
Probationers. Utah Division of Corrections.

Wisconsin Division of Corrections. Dept. of Health and Social
Services. (1986). Project Remrt #2: Development of the
Wisconsin Risk Assessment Scal_e.

Zhang, X., and DeLaPaz, C. (1990). The Evgluation of the Wisconsin
Qqssificgtion System as it Applies to The Los Angles
Probation Population. National Institute of Corrections.

"llllllllllllIll'llllllls