 

 

 

zo‘o LIBRARY
Michigan State
Ul liversity

 

 

 

This is to certify that the
thesis entitled

VALIDATION OF THE JUVENILE SEX OFFENDER
ASSESSMENT PROTOCOL (J SOAP-II) FOR
PREDICTING SEXUAL AND NON-SEXUAL

CRIMES

presented by
JODI PETERSEN

has been accepted towards fulﬁllment
of the requirements for the

MA. degree in f- .jislchologx
/ \

  
  
 
 

,/

///

, ”Major Professor’s Signature

f/éﬂ

 

Date

MSU is an Afﬁrmative Action/Equal Opportunity Employer

 

 

 

oo—---a-.---u-u-.-—--n--v----o-a-o---n---.-----.»

 

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProleoc&Pres/ClRCIDateDue.indd

 

 

 

VALIDATION OF THE JUVENILE SEX OFFENDER ASSESSMENT PROTOCOL
(J SOAP-II) FOR PREDICT ING SEXUAL AND NON-SEXUAL CRIMES

By

Jodi Petersen

A THESIS

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

MASTER OF ARTS
PSYCHOLOGY

2010

ABSTRACT

VALIDATION OF THE JUVENILE SEX OFFENDER ASSESSMENT PROTOCOL
(J SOAP-II) FOR PREDICTING SEXUAL AND NON-SEXUAL CRIMES

By

Jodi Petersen

Male adolescents are vastly overrepresented in rates of criminal sex offending
known to police (Barbaree & Marshall, 2008; United States Census Bureau, 2000).
Courts are then faced with the precarious position of determining the risk to the safety of
society that these youth pose and what type of treatment is most appropriate for them.
Risk assessment instruments are valuable in moving toward making these decisions based
on empirically derived risk factors and minimizing bias. The study reported here
assessed the predictive validity of one such measure, the Juvenile Sex Offender
Assessment Protocol (J SOAP-H, Prentky & Righthand, 2003), on a sample of male
adolescents (N=125) adjudicated for sex offenses in two Midwestern industrialized
county courts. Twelve month follow up time was utilized to assess the predictive ability
of the measure for both sexual and non-sexual recidivism. Additionally, risk group cut

offs (low, moderate, and high) were developed based on this population.

TABLE OF CONTENTS

LIST OF TABLES .............................................................................................. iv
LIST OF FIGURES ............................................................................................. v
INTRODUCTION .................................... ‘ ........................................................... 1
LITERATURE REVIEW ....................................................................................... 5
METHODS ...................................................................................................... 40
RESULTS ........................................................................................................ 46
DISCUSSION ................................................................................................... 50
REFERENCES ................................................................................................. 65

LIST OF TABLES

Table 1. Juvenile Sex Offender Risk Assessment Literature Summary ............................. 54
Table 2. Correlations Between J SOAP and Sexual Recidivism ...................................... 57
Table 3. Correlations Between J SOAP and Non-Sexual Recidivism ................................ 58
Table 4. Prentky Inspired Risk Groups ................................................................... 59
Table 5. ANOVA for Prentky Inspired Risk Groups ................................................... 60
Table 6. Empirical Risk Group Descriptives ............................................................ 62
Table 7. Signiﬁcant Items for Predicting Sexual Recidivism ......................................... 63

LIST OF FIGURES

Figure 1. Number of Youth at Each J SOAP Score ..................................................... 61

Introduction

Sex crimes occur in our communities on a daily basis. Adolescent boys account
for over one third of sex offenses reported to police that are committed against minors
(OJJDP, 2009) and are responsible for one ﬁfth of all rapes and one half of all child
molestations committed each year (Barbaree & Marshall, 2008). Boys under the age of
19 account for 28.6% of the total US population (United States Census Bureau, 2000),
making male youth over represented in the sex offending population.

Sex crimes appear to be concerning to the general public as evidenced by
research, frequent media coverage, and policy regarding these crimes. The Institute for
Social Research reported that over 88% of high school seniors worry about crime and
violence (2003). Overall prevalence of sex offense arrests have been decreasing, but for
juveniles who are 13 years old or younger, there appears to be an increasing trend
(Barbaree & Marshall, 2008).

As a result of public fear, sex crimes have also been the impetus for much
legislation. Since the early 19003 there have been initiatives to register sex offenders,
notify the community, and enforce residence restrictions (Lafond, 2005). More recent
legislation has included the Jacob Wetterling Crimes Against Children and Sexually
Violent Offender Registration Act in 1994 which requires all states to maintain a registry
of sex offenders’ addresses. This law was amended in 1996, allowing for community
notiﬁcation (Levenson, Brannon, Fortney, & Baker, 2007). This law has commonly
become known as “Megan’s Law”. This was a national law that was enforced and
maintained on a state level. The murder of 9-year-old Jessica Lunsford by a convicted
sex offender spurred “Jessica’s Law” in 2005 which increased criminal sanctions for sex

crimes against children (Levenson et a1., 2007). The Adam Walsh Child Protection and

1

Safety Act of 2006 established a national registry and allowed law enforcement to reach
across state boundaries in the pursuit and prosecution of sex offenders (Levenson et a1.,
2007). The Adam Walsh Act also included the Sex Offender Registration and
Notiﬁcation Act which provided minimum standards for sex offender registration and
community notiﬁcation across the US (United States Department of Justice SMART
Ofﬁce, 2007).

These laws serve to make the public aware of the sex offenders that live in their
communities. A survey of Washington residents in 1997 indicated that 80% of those
surveyed were familiar with notiﬁcation legislation such as Megan’s Law and 80% of
those people believed that the law was very important, making them feel safer by
knowing where convicted sex offenders lived (Levenson et a1., 2007).

Crime is perceived to be a large issue in our society, sex crimes especially. The
large amount of legislation regarding sex offenses and the public fear of sexual
victimization make this a social issue that merits concern not only for the safety (both
perceived and actual) of the community, but also for the control and rehabilitation of sex
offenders. There have long been ideas regarding why people commit crimes and why
certain types of people commit certain types of crimes. It seems however, that sex
offenders are not as well understood as other types of offenders, and juvenile sex
offenders are even less understood (Barabee & Marshall, 2008). This fear and lack of
understanding has lead local courts to look for ways to best assess J S03 and attempt to
predict their risk for committing future sex offenses (recidivism). While much of the
literature refers to juvenile sex offenders in general, most research on this topic is

actually done on male juvenile sex offenders. Therefore, this document will refer to the

extant research as being descriptive of male juvenile sex offenders, even when the
literature does not specify the gender of the sample being used, and not generalize such

information to female juvenile sex offenders, as they are not of the same population.

The study reported here adds to our understanding of male juvenile sex offenders’
proﬁles, unmet needs, and risks using a risk assessment measure known as the Juvenile
Sex Offender Assessment Protocol II (J SOAP II, Prentky & Righhand, 2003). It also
examined the ability of the J SOAP H to predict future crimes by male youth who have
been adjudicated for a sexual offense. The following sections review what is known
about male J 805 as a population, how they are being currently handled (legally and
through treatment), why risk assessment measures are necessary, what juvenile sex
offender risk assessments have been created and used, and why the J SOAP II is used in
this study. This will be followed by a statement of the rationale for this study, its

research questions, methods, results, and discussion.

The study reported here adds to the limited body of literature by expanding our
knowledge of male juvenile sex offenders and the utility of risk assessment measures
such as the Juvenile Sex Offender Assessment Protocol II. The study reported here also
adds to practitioners’ ability and conﬁdence in determining which male offenders are
most dangerous to the public (likely to recidivate) and which offenders pose a lower risk
for recidivism. Separating offenders by risk theoretically holds the promise that low risk
offenders can be kept in community-based, lower cost treatment programs, alleviating
funding that can then be focused on more intensive treatment for offenders who pose a

greater risk.

Literature Review

Male juvenile sex offenders (J SOs) are a sub-population of juveniles who have
committed crimes. While there are state deﬁned sex crimes, for the purposes of this
study, a sex oﬂender is actually labeled by the court and includes anyone who has been
adjudicated for committing the state-deﬁned sex crimes as well as anyone the court
deems as having a sex-related offense or needing sex-speciﬁc treatment. When
understanding male J $08 as a population, one must understand what types of offenses are
included in this population, demographics, patterns of behavior, personality factors which
indicate areas of risk, and circumstances of offending. The next section will examine
these issues, their link to recidivism, and the development of risk assessment measures.

This literature review was developed using various search engines. Proquest,
PsychInfo, and the National Criminal Justice Reference Service abstract database were all
searched using the terms (adolescent OR juvenile) AND (sex offender OR sex offense) in
the title or abstract. This brought about research on all areas applicable to this
population. Scholarly journals and dissertations were used while newspaper articles and
other sources were excluded. A meta-analysis of J SO risk assessment instruments is also
cited (Petersen, et a1., 2010) where the names of the three most common risk assessment
instruments were also searched and “ﬁle drawer searching” was utilized, contacting
authors of studies that were cited in other research but were not published. This made
available relevant research that has been presented at conferences or that has not been
accepted for publication. There were no date ranges used for either of these searches.
This literature search found a limited amount of published studies (but several other

theses/dissertations on related topics) as will be detailed further below.

Prior reviews of the literature on this population pointed out several areas needing
more exploration. One such review, although completed 17 years ago (Becker, Harris, &
Sales, 1993), found 73 articles on this population, 59% of which were merely
descriptions of the characteristics of the population, 31% were on treatment, and 10% (7
articles) were on miscellaneous other issues including assessment and recidivism. The
number of articles on assessment has grown in the past 17 years, but a recent review of
the J SO assessment literature (Peterson, 2010) found fewer than 30 studies before
employing “ﬁle drawer searching”, suggesting that more published, disseminated
research is needed and that there may be some sort of disconnect between doing research
and publishing research on juvenile sex offending. Again, the vast majority of these
studies focus on only male juvenile sex offenders. Caution should be used when
attempting to use such information to understand female juvenile sex offenders, as their
characteristics and etiology of offending may be very different from male juvenile sex
offenders. As will be seen in the review summarized below, the quality of extant
research in this area may indicate that publication can be a challenge. This review will
ﬁrst discuss the etiology of offending, including what juvenile sex offending by males is,
the typical male J SO demographics, patterns of behavior, personality patterns as related
to risk, the common circumstance of offending, and the current treatment modalities
used. After describing the population, the research on risk assessments will then be
detailed including, what they are, what elements should be included, what assessments
are available (including a description of the three most common measures: the Juvenile
Sex Offender Assessment Protocol, the Estimate of Risk of Adolescent Sexual Offense

Recidivism, and the Juvenile Sexual Offense Recidivism Rate Assessment Tool). After

describing the measures and comparing them, the rationale for using the chosen measure
will be detailed.

What is juvenile sex offending?

Generally, a sex offense is considered an act of sexual behavior that is
“exploitive, manipulative, aggressive, threatening, or without true consent” (Gibson &
Vandiver, 2008, p. 3). When dealing with juveniles this becomes complex, especially in
states such as Michigan, where no one under the age of 13 is considered legally able to
give consent to any sexual activities and no one under the age of 16 is considered legally
able to give consent to any act involving penetration (Michigan Penal Code, 2009).
These offenses can take on several legal titles depending on their severity. The most
common legal titles in the state of Michigan are criminal sexual conduct (ﬁrst through
fourth degrees), gross indecency, indecent exposure and lewd and lascivious conduct
(Michigan Penal Code, 2009). For some states there is a minimum age decided upon by
the state for a juvenile to become labeled a J SO; offenses committed by juveniles under
that age are considered to have sexual behavior problems, but are not typically
prosecuted. Michigan has no such age minimum. To gain a complete understanding of
male J 808, their common demographic traits, patterns of offending behavior, patterns of
risk for recidivism, and common circumstances of offending will be explained, as well as

the current treatment modalities being used.

Demographics

Male J 803 are a population who do not appear to be similar to adult sex offenders

or other types of juvenile offenders (Andrade, Vincent, & Saleh, 2006; van Wijk, van

Horn, Bullens, Bijleveld, & Doreleijers, 2005; Gibson & Vandiver, 2008). Van Wijk et
al. (2005) used information from psychological assessments on 600 male adolescent
offenders in the Netherlands who were deemed to be serious offenders by prosecutors.
They found that White youth were disproportionately represented in the male J SO
population while minorities were overrepresented in the population of other juvenile
offenders. Racial differences may exist in juvenile sex offending, as suggested by
McIntyre’s (2009) study using the Minnesota Multiphasic Personality Inventory (MMPI),
with male African American J 805 scoring signiﬁcantly lower on items on sexual abuse
victimization, physical abuse victimization, and psychopathic deviance than their

Caucasian counterparts.

There are notable gender differences in juvenile sex offenders. Excluding
prostitution, females only account for 7% of sex crime charges for youth (Snyder, 2002).
Since females are in the minority, they are often excluded from research on juvenile sex
offending in general, making research on female offenders extremely limited. Kubic,
Hecker, and Righthand (2002) did one such female I SO study on 11 youth, comparing
them to an age matched sample of 11 female adolescents with a history of non-sexual
person offenses. They found that the female sex offenders had signiﬁcantly fewer
behavior problems but had begun their offending behaviors at an earlier age than their
non—sex offending peers. This same study also compared the 11 female J $08 to an age
matched sample of 11 male sex offenders and found them to be extremely similar. The
only area where they appeared to be different was in history of abuse, where female
offenders had experienced signiﬁcantly more severe and pervasive abuse than their male

counterparts. Females are also typically younger and more likely to have a male victim

than male J SOs (Vandiver & Teske, 2006). Research has also shown that J 305 are
signiﬁcantly younger than other types juvenile offenders (e. g. van Wijk et al., 2002). For
these reasons, male and female J 803 should be studied separately, and extant knowledge
based on research, including risk assessments developed on male only or mixed gender

samples, should not be applied to female juvenile sex offenders.

Family characteristics play an important role in understanding male juvenile sex
offenders. Wieckowski, et al. (1998) found that youth in a residential treatment center
for sexual offending in Virginia typically came from “multiproblem families” with
histories of abuse in childhood, and exposure to pornographic materials. Caregiver
instability/deﬁciency and maltreatment have also been suggested to have an impact on
later sex offending by youth. Daversa (2005) employed the Multidimensional
Assessment of Sex and Aggression (MASA) on a sample of 329 J 805 and found that
physical and sexual abuse in early childhood were linked to socially incompetent
behavior and future sex offending in adolescence. Parent ratings using the Behavior
Assessment System for Children (BASC) do not appear to differentiate sex offending
populations from non—sex offending youth (Tomatis, 2007) although behavioral
differences are noted when using other data sources. The next section will explore the

common behavior patterns of J 805.
Patterns of Behavior

It is often suggested that male J 805 behave differently in social settings from
other youth or other types of male juvenile offenders. In a study of 32 male J 803 and 82

other types of offenders, Buttler and Seto (2002) found that J 805 were not signiﬁcantly

different from non-sex offenders in childhood conduct problems or current behavioral
adjustment or antisocial attitudes/behaviors. This same study found that youth who
committed sex offenses and noother types of offenses had fewer conduct problems,
better pro-social attitudes and current adjustment than non-sexual offenders. However,
another study showed that J 805 in a residential treatment setting had greater
psychological dysfunction and poorer social functioning than non-sex offenders in the
same residential setting (Leguizamo, 2000). The Denver Neighborhoods Study, based on
a sample of 78 J 803, 156 non-sex offenders, and 80 non-offenders, found that J 803
perceived themselves to be more isolated from peers, families, and schools than non-sex

offenders and non-offenders (Miner & Munns, 2005).

Empathy is thought to be an important personality characteristic to understand
when working with J 303, but research on this characteristic is lacking (Curwen, 2003).
Curwen (2003) used the Interpersonal Reactivity Index (IRl) to assess empathy in 1803
and found that sexual violence is related to empathy while diverting blame for the offense
and endorsing violence is related to discomfort in emotional situations. Curwen also
suggested that differences in empathy scores for J 803 may be more closely related to
victim characteristics or the speciﬁc circumstances of their offending. Using the same
scale, Lindsey, Carlozzi, and Eells (2001) found signiﬁcant differences between J 805,
non-sex offenders, and non-offending youth on personal distress (with both types of
offenders scoring higher than non-offenders) and empathic concern (with non-sexual
offenders scoring higher than sex offenders). A third study using this same measure (IRI)
also found that JSOs scored signiﬁcantly lower on empathy than non-offenders (Burke,

2001). Alternatively, Monto et al. (1994) found that there was no signiﬁcant difference

on an empathy questionnaire between J S05 and non-offenders in a study of 82 male J 805

and 108 male non-offenders.

There have also been many psychiatric diagnoses that are thought to explain male
juvenile sex offending throughout history. Across mental illnesses, fewer male J SOs
report a history of contact with a mental health practioner and also have fewer mental
illness diagnoses than general juvenile delinquents (e. g. Vandiver, 2006; Page, Tourigny,
& Renaud, 2009). However, male J 808 demonstrated more psychopathology than an
inpatient comparison group of youth with conduct disorders, substance abuse, major
depression, or Attention Deﬁcit Hyperactivity Disorder (ADI-ID) as evidenced by the
Minnesota Multiphasic Personality Inventory (MMPI); although it should be noted that
the study reported here had a small sample of 59 J 803 and 15 inpatients (Herkov,
Gynther, Thomas, & Myers, 1996). Additionally, in a recent study of 70 J 805 conﬁned a
residential treatment center, J 805 were found to be more likely to meet the Diagnostic
and Statistical Manual of Mental Disorders IV (DSM-IV) criteria for dissociative
disorders than a comparison group of 47 psychiatric inpatient youth (10 J SO youth
compared to 2 non-offending youth, no signiﬁcance tests were reported) (Friedrich, et al.,
2001). Positive correlations have been found between occurrences of Attention Deﬁcit
Hyperactivity Disorder (ADI-ID) and conduct disorders with violent sex crimes and rapes,
but not sexual offending by juveniles in general (V aih-Koch, Ponseti, & Bosinkski,
2001). Youth with conduct disorders, both J S03 and non-sex offenders, were compared
using the Million Adolescent Personality Inventory (MAPI) and found to have signiﬁcant
differences on social tolerance and social conformity subscales, indicating that the J SOs

were more passive, dependent on others, compulsive, socially isolated, awkward, and

10

uncomfortable with sexuality than non-sex offending youth with conduct disorders

(Maring, 1998).

In school settings, differences may also be seen between J 805, other offenders,
and non-offending youth. One study found that a larger proportion of J 308 attended
special education classes than non-sex offenders (van Wijk et al., 2005). Male juvenile
sex offenders and non-sex offenders were also found to have more school normlessness
than non-delinquents, with J 808 having more peer normlessness than non-sex offenders
and non-offenders in the Denver Neighborhood Study mentioned previously (Miner &
Munns, 2005). Although more school and peer normlessness was found, J 303 typically
have fewer ofﬁcial school-related behavioral issues than non-sex offending youth

(Tomatis, 2007).

Several taxonomies have been proposed to describe the behavior patterns of adult
sex offenders, but they do not appear to translate to J SOs. Thus, different taxonomies
have been proposed for J SOs. Burton (2000) said there are three categories, ‘early
offenders’, ‘teen offenders’, and ‘continuous offenders’, based on the age sex offending
behaviors begin and chronic behaviors versus isolated incidents of offending behaviors.
In addition to age differences, ‘continuous offenders’ had more contact, non-contact, and
penetrative acts than the other two groups and also had higher rates of being the victim of
physical and emotional abuse. Burton suggested that the pattern of risks for the

‘continuous offenders’ led them to be at a higher risk for recidivism.

Similarly, Mofﬁtt (1993) created a taxonomy of antisocial behavior which

suggested ‘life-course persistent’ and ‘adolescent limited’ groupings, based on her

11

hypothesis that youth with early disruptive behavior patterns have greater criminal
careers both in severity and length. This taxonomy has been supported in later studies
looking at general crime, but this has not been replicated with 1805 (Andrade et a1.,

2006)

Another study used cluster analysis (Pithers, Gray, Busconi, & Houchens, 1998)
to identify ﬁve subgroups of J 805. These groups were labeled ‘nondisordered’, ‘abusive
reactive’, ‘highly traumatized’, ‘rule breakers’, and ‘sexually aggressive’ based on
demographic characteristics, histories of abuse, psychiatric diagnoses, offense
characteristics, and behavioral checklists. The sexually aggressive group had the greatest

number of aggressive sexual acts and was most likely to recidivate into adulthood.

Hunter, Figueredo, Malamuth, and Becker (2003) proposed another typology for
JSOs, dividing them into two groups: those who abused prepubescent children (under age
12) and those who abused pubescent and post-pubescent females. The youth who
offended against prepubescent children were less aggressive, more likely to victimize a
family member, and had greater psychosocial deﬁcits. The study reported here used a
small sample size (less than 200) and did not account for youth who offended against

pubescent or post-pubescent males.

While these taxonomies appear to aid in our understanding of J 805, they are
relatively new, with none found developed prior to 1993. They are limited in their
empirical support due to small sample sizes and limited replication. Burton’s (2000)
groups regarding time period of initial offense were formed based on three groups (two in

residential facilities, one in a half way house) in Michigan with 122 adjudicated juvenile

12

 

sex offenders and 150 youth adjudicated for other offenses who reported sex offending
behavior at some point in their lives. Mofﬁtt’s taxonomy (1993) is based on reviews of
literature and basically groups youth into those who recidivate (“life-course persistent
offenders”) and those who do not recidivate (“adolescent limited offenders”). Pithers and
colleagues ( 1998) proposed a taxonomy using a sample of 127 children ages 6 to 12
whose means of referral were unknown. The sample was 65% male youth and 35%
females. Pither’s study used interviews from these children and their caregivers, but their
ofﬁcial offending status was not reported. Hunter et a1 (2003) created a taxonomy based
on the type of person offended against based on a sample of 206 youth from multiple
public and private residential, mental-health-oriented treatment centers. Only Mofﬁtt’s
taxonomy (1993) attempted to reach beyond convenience sampling of unrepresentative
samples. There is no agreement across these taxonomies about how to characterize J 803
since they range in their categorizations. These taxonomies aid in our understanding of
who J 808 are, but they have limited interpretability since we cannot use the taxonomies
to aid greatly in treatment planning or to predict sexual recidivism. Simply
understanding the population is not enough to implement productive treatment and keep
the community safe. Thus, while these taxonomies are helpﬁrl, they do not solve the
issues that the study reported here attempts to address, namely which offenders are most
needing of treatment and resources. The common behaviors of J 808 can be categorized
by such taxonomies. Along with behavioral patterns come personality patterns. The next
section will detail these common personality patterns and how they relate to criminogenic

risk.

13

Personality Patterns and Risk

Additional behaviors noted as possibly being risk factors for committing future
sex offenses include pervasive anger, history of conduct disorder, school behavior
problems, and antisocial behaviors (Prentky & Righthand, 2003). There are also several
attitude based risk factors regarding the youth’s understanding and reaction to his or her
sex offense (Prentky & Righthand, 2003). Again, across the literature there have been

few conclusive studies that result in agreement of these factors as being indicators of risk.

In regards to personality characteristics, van Wijk et al. (2005) found that violent
sex offenders were signiﬁcantly more extraverted and impulsive as well as having higher
‘lack of conscience’ and neuroticism scores than other offenders. Sex offenders also had
higher IQ scores than non-sex offenders and had lower school drop-out rates but had
signiﬁcantly more problems with peers. There were no signiﬁcant differences in scores
of bullying victimization or self esteem (van Wijk et a1., 2005). Also, victims were likely
to be family members, especially when the victim was younger than the offender (Gibson

& Vandiver, 2008).

While male J 305 appear to be a very heterogeneous population, there are some
common trends in personality and contextual characteristics. They tend to be socially
isolated from their peers and are described as loners who lack social skills necessary to
develop relationships (Ford & Linney, 1995). They tend to be more shy, timid, and
withdrawn, and are typically more behaviorally compliant than other types of offenders

(DeNatale, 1989). Family dysﬁmction and violence also seems to be prevalent among

14

this type of offenders (Blaskeet et al., 1989). Miner and Munns (2005) found that male
juvenile offenders (both general and sexual offenders) rated higher on a measure of
school normlessness than non-offenders, using a Likert style scale which measured
willingness to deviate from norms in an effort to maintain parental acceptance. They also
found that male J 808 rated their levels of perceived family and social isolation
signiﬁcantly higher than both general juvenile offenders and non-offenders. This
indicated that these feelings of isolation and normlessness may be particularly
characteristic of this population. Several theories of sex offending have called upon this
lack of social support and increased isolation and being a possible cause of sex offending,
as many sex offenses appear to be efforts to exert social control and power (Erooga and
Masson, 2006; Terry, 2006). Again, while past research has aided in our understanding
of the characteristics of JSOs, the research still lacks much utility in that it is difﬁcult to
use this information to predict recidivism or to gear individual treatment plans to target
changes toward the most “risky” factors. The study reported here addresses this
disconnect by using a risk assessment measure to predict recidivism and direct more
intensive treatment to those who need it most.

While understanding the population is important, the implications of this
information are limited without tying the knowledge to topics such as treatment and
recidivism. Although there are personality and behavior differences across offenders,
there are some notable similarities or patterns across offending. Of the studies analyzing
personality factors, only the 14 reported in Table 1 tie those factors to recidivism. There

is still much to be known about how personality is related to risk for reoffense. An

15

additional area of interest that can aid to our understanding of J 803 is their circumstances
of offending, as discussed in the next section.

Circumstances of Offending

There are several ways that a youth can become labeled a sex offender.
According to Gibson and Vandiver (2008), they can be categorized into three
circumstances. The ﬁrst of these consists juveniles engaged in “consensual” sexual
activity—youth who are agreeing to sexual activity but who are seen as unable to consent
in the eyes of the law due to age or mental ability. The second group is ﬁrst time
offenders who use force/coercion; this is suspected to be the largest group that becomes
known to the public/law enforcement. The third group consists of repeat abusive J 808 ;
it’s estimated that this is only 4 to 8 percent of all J $03, a much smaller number than the

public would generally presume.

Seventy percent of all sex offenses committed by adolescent boys are conunitted
in the home of the victim, offender, or another person rather than in a public setting and

are most common between 3:00 and 6:00 P.M. (Gibson & Vandiver, 2008). No studies

were found linking the time or location of sex offending by juveniles to recidivism.

While we can use past research to categorize offenders, the literature is still
lacking in understanding how these circumstance relate to recidivism. The study reported
here served to link what is understood about male sex offenders to treatment
severity/modality by identifying the offenders who are most in need of the greatest
intervention. In order to understand different possibilities in treatment, understanding the

current treatment modalities being used is essential.

16

The Treatment Status Quo

There are currently several trends in treatment for J SOs in the United States.
There is a severe lack of high quality research (especially random control trials), much of
the treatment that is currently used is based on strategies that are simply thought to be
beneﬁcial, as this section will describe. The modality of treatment ranges ﬁom
individual therapy, to court-run programs, to private community run programs (where the
youth continues to live in the community), to secure lock-down residential facilities.
Many local agencies develop their own treatment programs on a county or state level.
Regardless of modality, treatment programs currently tend to have similar themes such as
relapse prevention, reducing denial, increasing accountability for offenses, increasing
understanding of the impact of such offenses, understanding motives, sexual triggers and
risk situations, education about sexuality and norms, eliminating deviant arousal,
cognitive restructuring around deviant beliefs/attitudes, social and interpersonal skill
building, anger management, and family therapy (Davis & Leitenberg, 1987; McMackin,
et al., 2002). Exactly what these descriptors mean is difﬁcult to discern. A meta-
analysis of sex offender treatment programs (not differentiating between juveniles and
adults or specifying gender) found 12 studies of treatment versus comparison groups with
1,313 offenders and a recidivism rate of 19% with the treatment group versus 27% with
the non-treated group (N agayama Hall, 1995). They found larger effect sizes in studies
with follow-up lengths of greater than 5 years, those who included outpatients, and
involved cognitive-behavioral or hormonal treatment elements (rather than behavioral
treatment). They noted that the 2 studies of the 12 they reviewed with the greatest effect

sizes were also those with JSO samples but noted that these studies had relatively short

17

follow up times (actual length was not reported). An additional review of sex offender
treatment programs encompassing 80 comparisons and more than 22,000 people included
7 studies with samples of 1805 and again found higher effect sizes than programs treating
adult sex offenders (Losel & Schmucker, 2005). When not differentiating between adult
and juvenile sex offenders, they found a mean odds ratio of 1.70 (within the typical range
for offender treatment meta-analyses), that sex offense speciﬁc programs have a greater
impact than general programs on sex offenders, well structured cognitive behavioral
programs and voluntary treatment seem to be most related to reduced recidivism rates.
They also noted that only 40% of the studies had a score of 3 or higher on the Maryland
Scale of Methodological Rigor, a 5 point scale that judges quality based on
methodological features which could lead to causal interpretations, indicating only
moderate quality, and found the same limitations noted here, with nearly half of the
research taking place outside of the United States, small sample sizes, short follow-up
times, and limited details about the sample. A score of 5 on the Maryland Scale of
Methodological Rigor describes pure random designs, while a 4 describes studies with
procedures to ensure group equivalence, and the standards decrease to a score of l,
uncontrolled studies. Sapp and Vaughn (1990) found in a nationwide survey of
correctional facilities with programs for JSOs that there were 338 different therapies and
techniques being used, many of which relied on psychological therapy and behavioral
modiﬁcation. In that same study, the directors of the surveyed programs reported that
they would like to add 190 additional treatment strategies (totaling 528 types). This
averages to a desired 17.6 different types of therapy for each program in the study.

Additionally, a review of the research reported that the primary strategy of current

18

treatment programs consists of cognitive-behavioral techniques, psycho-educational and
pharmacological interventions (Ef’ta-Breitbach & Freeman, 2004). These programs also
often include individual, family, and group settings. A survey of the directors of 30 J SO
treatment programs in state run institutions found that programs had a mean length of
17.5 months and over half of the programs were mandatory (17 pro grams); only 1
program used a biological model, employing injections as treatment (Sapp & Vaughn,
1990). This same study found that 70% of directors said their best hope was to lengthen

the time between offenses and reduce the seriousness of future offenses.

In line with the meta-analyses reported above (Losel & Schmucker, 2005;
Nagayama Hall, 1995), the majority of the research found on J SO treatment focuses on
one speciﬁc setting and describes the treatment modality, as will be described in the next
part of this section. Very rarely are these studies tied to recidivism data. A community-
based step down model for J SO treatment using multi-modal therapy and an educational
focus has been suggested, and outlined in research by Casines (2003). Other pro grams
are designed for speciﬁc subgroups of J 808, such as those who have been victims of
sexual abuse themselves (Lindmeier, 2002; Woods, 1997), aiming to treat what they see
as the true root of the deviant behavior. Alternative treatment modalities are also

emerging, utilizing techniques such as music-centered creative arts therapy programs

(Skaggs, 1997).

A meta-analysis was done assessing the effectiveness of treatment for J 803 using
recidivism as an outcome (Reitzel & Carbonell, 2006), ﬁnding only 9 studies that
measured effectiveness by recidivism. These studies used an average 59 month follow up
(just less than 5 years). Four of the nine studies used a control group and the other ﬁve

19

used a comparison group. There was an average weighted effect size of .43 (CI = .33-
.55), suggesting statistically signiﬁcant treatment effects for sexual recidivism. The
author suggested caution for interpretation due to individual study characteristics such as
non-equivalent follow-up periods and the handling of dropouts. A longitudinal follow up
study of 100 J $05 in a residential treatment program stands as an exemplar for assessing
whether or not the treatment program met the pre-set clinical goals after release
(Eastman, 2005). More studies such as this will move the ﬁeld toward accountability and
greater understanding regarding what goals are realistically being met. Additionally,
research needs to go a step further and look at recidivism to see if the goals being met
have any impact on reducing recidivism. The nine studies from the meta-analysis
discussed above are simply not enough to adequately determine the effectiveness of J SO
treatment. Comparisons between types of treatment programs (modality differences as

well as multi-modal residential vs. community based) are needed.

There have been some published studies that look at the effect of certain
characteristics on treatment outcome. Factors such as age, extent of sexual knowledge,
maladjustment, impulsivity, defensiveness, prosecution, cognitive distortions, and
preoccupation are often used to predict treatment outcome (e. g. Kraemer, Salisbury, &
Spiehnan, 1998; Eastman, 2005; Anaforian, 2009). Cognitive distortions and some
background variables (intellectual functioning, history of victimization, history of
witnessing domestic violence) have been shown to be able to differentiate between youth
who complete treatment and those who never enter treatment or do not successfully
complete treatment (Eastman, 2005). Additionally, program criteria have been assessed

in relationship to the length of stay in a treatment program, with greater years of staff

20

experience being signiﬁcantly related to extended stays while sentence type, number of
offenders, funding type, exit criteria, and offense were not signiﬁcantly related to stay

length (Cooper, 1997).

There have been several calls for more research on treatment and increased
training to develop specialized treatment professionals (e. g. Perry & Orchard, 1992;
Epps, 1994). It is suggested that treatment practitioners need increased understanding of
the population, the myths, and the self-examination necessary to work with this
population (Perry & Orchard, 1992). Residential treatment facilities which typically
house more aggressive youth may require additional training for staff regarding the
management and treatment of sexually aggressive behavior (Epps, 1994). Training on
risk assessments could also be beneﬁcial in helping practioners to better understand the
J SO population. The variety of treatment programs available and the limited research on
the effectiveness of these programs points to the need for better individualized
understanding of J 805. This is one of the many reasons why risk assessment instruments

become vital, as will be described in the next section.
Are risk assessments needed?

In order to best treat and understand the male adolescent, risk assessment
measures have been shown to be useful tools in making decisions. Risk assessment
measures provide court workers with research-based tools to assist in decision making
regarding placement of juveniles (community-based or residential, depending on what
level of protection is needed for society and how intensive the needs/risks of the juvenile

are) as well as treatment planning. It is suggested in the literature that I SO risk

21

assessments should contain a comprehensive assessment of individual and contextual
factors (Grisso, 1996; Borum & Verhaagen, 2006), including unique risk factors for each
individual, with an understanding of the low base rate for reoffending. Both static and
dynamic risk factors should be included in order to get a ﬁrll picture of the needs and risk
presented by a J SO (V itacco, et al., 2009). Again, measures not developed speciﬁcally

for female juvenile sex offenders should not be used to assess the risk of females.

For court purposes, risk assessment measures aim to help in the decision making
process by using past data to help predict future crime (Funk, 1999). Risk assessment
measures are also used to provide structure, reliability, a better understanding of unmet
needs, and aid in creating treatment plans (Schmidt, Hoge, & Gomes, 2005). There have
been three generations, or types, of risk assessment measures. First generation measures
are based on professional opinions and were not shown to be very predictive of
recidivism (Bonta, 1996). Second generation measures consist of static predictors, which
cannot change with treatment (such as offense history, past behavior, etc) (Schwalbe,
2007). Third generation risk assessments include both static and dynamic factors
(dynamic factors are things that can change such as empathy, school stability, substance

use/abuse, support systems, etc) (Schwalbe, 2007).

The National Council on Crime and Delinquency (1998) set forth standards,
suggesting that risk assessment measures be tested for (1) validity, (2) reliability, (3)
fairness across all types of juvenile offenders, and (4) should also be useful for court
workers. For offense speciﬁc measures such as the J SOAP II, fairness across all types of
juvenile offenders refers to differences in age, gender, race, etc, rather than offense type.

Again, gender must be considered careﬁrlly, as male and female JSOs are not of the same

22

population. Therefore, meeting the above described standards includes not applying risk
assessment measures to females if they were not designed or validated speciﬁcally for
females. While most measures, including the J SOAP II, have some evidence for all four
of these factors when they are published, more evidence is needed to ensure that the
measure is truly useful. Many of the J SO risk assessment measures available were
created on small, unrepresentative samples. The generalizability and overall utility of
these measures is then unclear. Table 1 outlines the literature that could be found on risk
assessment measures for juvenile sex offenders. Again, these studies were found via a
thorough search of the extant literature in the ProQuest, PsychInfo, NCJRS, OJJDP, and
MedLine search engines using the search terms (adolescent OR juvenile) AND (sex
offender OR sex offense). As can be seen from this table, the J SOAP I has 4 published
studies, the J SOAP H has 6, the ERASOR has 2, the J-SORRAT has 2, and any other
developed measures have 1 or less study testing the validity or reliability of the measure
(the ﬁrst two standards mentioned above). The methodological shortcomings of this

research will be discussed in ﬁuther detail in the next section.

As Table 1 shows, research on this area is extremely limited. Of the 14 studies
reported in this table just over half report a follow-up length, ranging from 12 months to
13 years. In none of these studies with differential follow-up times was time at risk
accounted for. None of these studies use the risk assessment measure in representative
settings, such as a county court. All of them use samples that have already been sorted,
with some youth in residential treatment, or youth who have not all been adjudicated for a
sex offense, as intended by the measures. All studies omitted female offenders did not

account for gender differences.

23

The National Council on Crime and Delinquency’s third standard set forth,
fairness across all types of 1803, has not been fully tested in the existing research.
Stevenson et al. (2009) showed that registration was recommended more often when the
offender was Black, the victim was white (both regardless of the other party’s race), or
when the offender and victim were of different races. They also found that motivations
for registration were often to punish the offender, not necessarily to protect society.
These types of biases show a need for empirically proven risk assessment measures to aid
in decision making rather than relying on public opinions such as these that may also be

present in counsel, juries, or judges.

There are many risk assessment measures developed for predicting general
juvenile recidivism (Schwalbe, 2007). This is, predicting whether or not a juvenile who
has committed a non-sex related offense will commit another non-sex related offense in
the future. These general risk assessment measures were not created to predict speciﬁc
types of crimes such as sex crimes. Thus, sex offense speciﬁc measures are necessary, so
that they can focus in on this speciﬁc type of offender. There is a smaller pool of
possible risk assessment measures that are evidenced in the literature to predict future
sexual offending for male juveniles who have committed a sex offense. While most
courts see the utility of risk assessment measures and desire to use research in providing
the best treatment for youth that they are capable of, there are a multitude of possible risk
assessment measures available. While the utility of risk assessment measures is common
across types of juvenile offenders, there are several items or types of items that are
related to criminogenic risk speciﬁcally for male juvenile sex offenders, as will be

detailed in the next section.

24

What should be included in a Male Juvenile Sex Offender risk assessment?

Juvenile sex offender risk assessments for males can contain a variety of different
factors or items intended to predict risk. Previous non-sex offender speciﬁc risk
assessment literature has suggested that measures intended to predict recidivism should
have both static (historical, unchanging) and dynamic (treatment focused, changeable)
factors (Bonta, 1996; Schwalbe, 2007). A meta-analysis of 13 studies found 32 variables
that were used to predict sexual recidivism, only ﬁve of which were signiﬁcantly
predictive (Redlak, 2003). Those ﬁve items were: history of sexual abuse, adverse
family climate, out-of-home placement, deviant sexuality, and having more than one
victim. Another study looked at 86 residents of corrections-based juvenile sex offender
treatment program (not specifying any genders or analyzing gender differences) and,
using a self-created collection of assessment items, found higher recidivism rates for
youth who scored higher on impulsivity, had a male victim, higher numbers of prior sex
offenses, and present paraphilias (Miner, 2002). This study noted that signiﬁcant items
were not consistent across youth and that these items were not reﬂective of the literature
regarding important factors for predicting recidivism in adult sex offenders. A third
study used a relatively small sample of 25 youth with a three year follow up and found
that youth who were younger, had 2+ clinical diagnoses, had altercations (both physical
and verbal) while in treatment, and were in treatment longer had higher sexual recidivism
rates (Clybum, 2002). There were no items that were common across all three of these
studies. As these studies make clear, determining what items are signiﬁcant predictors of

sexual recidivism is still in debate.

What JSO risk assessments are available?

25

There are several J SO speciﬁc risk assessment measures that are presented in the
literature. Older measures typically include only static factors (second generation
measures) while newer measures typically include both static and dynamic factors (third
generation measures). The available measures also have varying levels of research to
support them and were initially validated on greatly differing sample sizes, as will be
explained in the following sections, using Table 1. There are three primary measures that
have been reported in the literature. These measures each have a minimum of two
empirical studies published. They are the J SOAP H, the ERASOR, and the J SORRAT II.
Each measure will be outlined below, including the measures’ characteristics and
supporting validation data. As the literature review will reveal, the small number of
measures and limited amount of validation data available point to the need for further
research in this area. The study reported here will add to the existing literature by
demonstrating the validity and utility of one measure in a sample that is quite different
than those where the measures were created. The study reported here uses the four
criteria proposed by the National Council of Crime and Delinquency (1998) to assess the
rigor and utility of one measure. By showing how this measure ﬁts the criteria, it builds
further conﬁdence for the measure to be used more widely, ﬁrrther reducing biases and

possibly improving outcomes for JSOs.

J SOAP.

The Juvenile Sex Offender Assessment Protocol II (J SOAP II) is a revised version
of a previous instrument and appears to be fairly widely used (Burton, Smith-Darden, &

Frankel, 2006). The J SOAP I and 11 combined have more empirical studies available

26

than other J SO risk assessment measures. This can be seen in Table l, which compares

the data available for all measures.

The J SOAP 11 consists of 28 items with 4 subscales including both static and
dynamic factors (making it a third generation measure) (Prentky & Righthand, 2003).
These scales are sexual drive/sexual preoccupation (static), impulsive/antisocial behavior
(static), clinical/treatment (dynamic) and community adjustment (dynamic). Items are
scored 0, for not present, 1, for suggested or partial presence, and 2, for clearly or always

present. There are no set risk-level cutoffs for the J SOAP II.

The original J SOAP was created using a sample of 96 youth, but found that the
recidivism rate and sample size were too small to make any analyses meaningful. While
the creators had several studies validating the original instrument which pointed to the
changes necessary which formed the J SOAP II, validation studies on the J SOAP 11 came
from outside researchers and are limited. Other studies on the J SOAP I included a study
of 253 “very hi gh-risk juvenile sex offenders” which found correlations between some
subscales and recidivism (Waite, Pinkerton, Wieckowski, McGarvey & Brown, 2002).
This study was not reﬂective of how the measure would typically be used since youth
were already sorted prior to assessment, limiting interpretability. An additional study in a
more reﬂective setting, but with a smaller sample size of 54 adjudicated male J 805,
found an AUC of .79 for total score predicting sexual recidivism (Hecker, Scoular,
Righthand, & Nangle, 2002). A later study using the J -SOAP I demonstrated an
intraclass correlation (ICC) of .70 and internal consistency (a) of .87 using archival ﬁle
coding for 60 male youth adjudicated for a sexual offense (Martinez, Flores, &

Rosenfeld, 2007). This same study found correlation coefﬁcients of .34 for total score

27

and any recidivism (p<.05), .31 for total score and sexual recidivism (p<.05), .33 for the
dynamic scales score and any recidivism (p<.05), .42 for the dynamic scales score and
sexual recidivism (p<.05), .13 for the static scales score and sexual recidivism (NS), and
.26 for the static scales score and any recidivism (p<.05). Also ROC AUCs of .76 for any
recidivism (p<.05) and .78 for sexual recidivism (p<.05) were found. This study was in a

treatment program in a mostly minority urban setting.

The J SOAP II has a growing body of literature, with 5 published studies found
using the measure to predict sexual recidivism. Additional studies have been done using
the measure to predict treatment outcomes or building the validity using factor analyses
and within-measure tests. The predictive validity studies on the J SOAP II are plagued
with the same issues of non-representative and small samples. Parks and Bard (2004)
used a sample of 156 male J 805 in a secure residential treatment facility found that the
J SOAP H was able to differentiate between types of recidivist (sexual versus non-sexual)
and that one scale (Impulsive/Antisocial behavior) was predictive of sexual recidivism.
While the Parks and Bard study is not extremely small, it is an unrepresentative sample of
youth who had been selected to attend a secure residential treatment setting.
Additionally, Mccoy (2008) used a sample of 128 youth in an outpatient treatment center,
ﬁnding some correlated subscales but no signiﬁcant prediction. Both of these samples
use a pre-sorted group of youth that are not reﬂective of how the measure would be used
in a prospective manner, the manner that is most useful to court and treatment workers.

A published grant update by the creators of the measure used a larger sample size of
nearly 800 youth with a longer follow-up period of between 6 months and 7 years

(Prentky, 2006). This study found an AUC of .824 for pre-adolescents and .803 for

28

adolescents for predicting sexual recidivism. This study was also done using an
unrealistic sample of youth in custody of the social service system, not youth adjudicated
for a sex offense receiving assessment at court intake. A study using the J SOAP H as a

comparison to other measures was also done and will be detailed in an upcoming section.
ERASOR.

The Estimate of Risk of Adolescent Sexual Offense Recidivism (ERASOR) is
another new and promising risk assessment measure for J SOs (Worling & Curwen,
2001). It is a 25 factor measure (items are referred to as factors for this measure)
intended to predict recidivism for individuals who are 12-18 years old. It contains mostly
dynamic (16 of 25) factors but also includes some static factors. The heavy emphasis on
dynamic factors makes this measure intended to predict short term risk of less than one
year, not long-term risk. There are ﬁve categories of factors on the ERASOR: Sexual
interests, attitudes and behaviors, Historical sexual assaults, Psychosocial functioning,
Family/environmental functioning, and Treatment. Each factor is coded as Present,
Possibly or partially present, Not present, or Unknown. These coding categories do not
result in a ﬁnal number score, but instead the creators suggest using the combination of
high risk categories (those with several present or possibly/partially present factors) as an

indication of overall risk.

Data on which the ERASOR was created were not provided with the initial
publication of the measure. The creation of the measure was based mostly on an
intuitive, iterative process between workers at the Sexual Abuse: Family Education &

Treatment Program (SAFE-T). Follow up data were published separately using a sample

29

of 136 male adolescents who acknowledged or were charged for a sex offense and who
were assessed at one of 45 different community-based agencies or one of 91 residential
treatment centers Morling & Curwen, 2004). Risk assessments were done at one of
three different points in treatment: intake, mid-way through, or discharge. The ERASOR,
as previously mentioned, does not give a ﬁnal number score, so the coder were instructed
to use the measure and their interpretation of the high risk factors to assign an overall risk
rating to the youth of low, moderate, or high after their clinical interviews with each
youth. ROC AUCs were calculated for total score and sexual recidivism at .72 (p<.05)
and overall risk estimate and sexual recidivism was found to be .66 (p<.05). It should be
noted that whether or not the youth recidivated was known at the time of the assessment
and was determined by whether or not the offense bringing the youth in for assessment
was the ﬁrst time he was caught by an adult (non-recidivist) or if he had been caught
committing a previous sex offense and then committed another sex offense that resulted

in the assessment (recidivist).
J SORRAT.

The Juvenile Sexual Offense Recidivism Rate Assessment Tool (J SORRAT H) is
a 12 item measure designed for male offenders ages 12 to 18 (Epperson et a1, 2005). It is
meant to be an objective measure where some items are scored 0 or 1 depending on the
presence of the factor, and some items are scored 0 to 2 or O to 3 to reﬂect the severity of
that risk factor. This measure was created on a sample of 636 males who were
adjudicated for sex offenses. They found an AUC of .89 for sexual recidivism as a
juvenile and .79 for sexual recidivism as an adult or juvenile. The data reported are from

an unpublished manuscript and no outside research could be found on this measure to

30

date except for one study comparing the measure to the J SOAP H, detailed in the next

section (V iljoen, et al., 2008).
Comparing Measures.

Few studies have been published which compare multiple J SO risk assessments
on the same sample of youth. Vilijoen et al (2008) compared the J SOAP H, J SORRAT
II, and SAVRY using a sample of 169 male youth in a residential treatment program for
J 805. It should be noted that this sample may not be reﬂective of all youth who are
adjudicated in a given county for a sex offense since they were somehow deemed in need
of residential treatment. The SAVRY (Structured Assessment of Violence Risk in
Youth) was developed for general juvenile offenders, not J S08, and thus will not be
discussed in detail in this document. This study used a 250 day follow-up period and risk
assessments were coded based on archival ﬁles. They found that the J SOAP H total score
had an AUC of .54 (non-signiﬁcant, NS) for predicting sexual recidivism, .56 for
predicting nonsexual violent recidivism (NS), .63 for predicting “serious” nonsexual
violent recidivism (p<.05), and .56 for predicting any recidivism (NS). The J SORRAT H
had an AUC of .53 for sexual recidivism (NS), .56 for nonsexual violent recidivism (NS),
.55 for “serious” nonsexual violent recidivism (NS), and .54 for predicting any recidivism
(NS). The SAVRY had an AUC of .53 for sexual recidivism (NS), .58 for nonsexual
violent recidivism (NS), .69 for “serious” nonsexual violent recidivism (p<.05), and .58
for predicting any recidivism (NS). None of the measures’ total scores or subscale scores
had statistically signiﬁcant AUC for the prediction of sexual recidivism in this study. It
should be noted that this study was also completed using an unrepresentative sample of

youth in a residential treatment program.

31

Why use the JSOAP II?

The J SOAP was originally created in 1994, making it the J SO risk assessment
measure with the longest history (Prentky & Righthand, 2003). Although much more
work is needed, it has more published validation work (measured by number of studies of
better quality, although the quality of these studies is still lacking) than any other
measures available and that validation work appears to be quite promising. The measure
is also free, making it a reasonable choice for courts and treatment centers. The J SOAP
II was chosen in this situation due to these factors as well as its arnenability to coding

from archival ﬁles.

As previously mentioned, the measure was constructed using a sample of 96 male
J 808 (ages 9 to 20, average of 14) who were referred to the institute for treatment, using
the measure at intake and again after 24 months of treatment (Prentky & Righthand,
2003). A twelve month follow-up on 75 of the youth (78%) showed an 11% total
recidivism rate (5 nonsexual offenses, and 3 sexual offenses, resulting in a 4% sexual
recidivism rate). The average total J SOAP score for the boys who did not reoffend was
21 and 30 for those who did reoffend sexually. However, due to this small sample, the

authors caution that results were not interpretable.

A follow-up study was done using treatment outcomes as the dependent variable
which pointed to several changes that required changes in the original J SOAP (Prentky &
Righthand, 2003). The revised scale, called the J SOAP H, was completed in 1998 and
contained 28 items with the same 4 subscales (maintained via a principal component

analysis). This measure was validated on a sample of 153 male J $05 with an average age

32

of 16 (Righthand, Prentky, Hecker, Carpenter, & Nangle, 2005). Internal consistency
was high for scales 2, 3, and 4 (with alphas of .88, .95, and .80 respectively) and
moderate for scale 1 with an alpha of .64. Concurrent validity was tested using
correlation between the J SOAP H and the Youth Level of Service/Case Management
Inventory (YLS/CMI), a general delinquency risk assessment measure, resulting in a

correlation coefﬁcient of .91 for total score.

While there are some follow-up studies, all of them are either on small sample
sizes (fewer than 100) or do not have fully representative samples of J 805, or both. In
order to fully understand the utility of the J SOAP H, it needs further validation studies on
larger sample sizes, samples from more geographic locations and also on the full

spectrum of JSOs that come into contact with the court.

The primary concern with many of these studies is that the sampling procedures
do not match how the measure would be used as a predictive decision making tool.
Many of the studies were on samples of all male youth in residential treatment facilities
(e. g. Martinez, Flores, & Rosenfeld, 2007; Prentky, Harris, Frizzell, & Righthand, 2000).
Measures such as the J SOAP H are intended to predict recidivism and can have a great
impact by aiding the court in determining how youth are treated. Relying on professional
judgment (coupled with fear of public repercussions for keeping a convicted I SO in the
community who then reoffends), the courts in this study state that many juveniles are sent
to residential treatment facilities when they may have been able to succeed in a
community based treatment program. Studies testing the predictive validity of measures

such as the J SOAP H should be conducted in the situations where the measure would be

33

used: based on information available after a court intake. This information could then

affect treatment modality.

An additional concern with many of the studies in Table l is the small sample
sizes and recidivism rates which lead to results that are not interpretable and/or not
generalizable. Some authors note that the results should be taken in caution due to the
small samples and low base rates (Prentky, Harris, Frizzell, & Righthand, 2000; Mccoy,
2008). Several other studies noted in Table 1 relied upon pre-sorted samples of youth
(e. g. Pinkerton, Wieckowski, McGarvey, & Brown, 2002; Parks & Bard, 2004; Hersant,
2007; Worling & Curwen, 2004). These pre-sorted samples may minimize the
differences in the youth and may also be a higher risk sample (those deemed needing
secure or residential treatment rather than community based treatment). This may limit
the predictive ability of the measures as well as the generalizability of these studies to
more realistic settings, such as the one in the study reported here. While correlations
between overall score and sexual recidivism rate hint to the validity of a measure,
receiver operating characteristic area under the curve analyses are more interpretable and
provide more conﬁdence in the validity of the measure. All previous studies using this
measure lack applicability to real life settings due to their pre-sorted samples. In order
for a court or “ﬁont door” setting to use the measure, they need to know about the
predictive validity of the measure for every type of youth that they see. Studies of other
samples of youth are beneﬁcial in adding to our knowledge, but do not add to the utility

of the measure.

Utility of the Current Study

34

The current study adds to the existing literature and may aid in the local treatment
of J 808 in several ways. The literature on J 808 and such risk assessments is very
limited, as previously discussed, with only 14 studies found using the three most common
J SO risk assessment instruments. What is available is limited both in number and scope,
with limited sample sizes, varying follow-up times, and unrepresentative samples, as
described above and detailed in Table l. The study reported here will add to the body of
literature by using a relatively large sample size in a location that is different than
previous areas studied. It will also utilize the J SOAP H in the way that it was intended to
be used and would likely be used by court systems. The study reported here uses a
sample reﬂective of all boys who enter the court system for committing a sex related
offense. This is not a study of a captive, unrealistic population in a residential treatment
center. It is a study of the full spectrum of youth who are charged with sex offenses. The
study reported here responds to the need in these settings for a measure to assess risk for
recidivism. Pre—sorted samples cannot possibly achieve this goal. No studies to date
have been published utilizing the J SOAP H with a representative sample of greater than

60 youth (as referenced in Table 1).

Additionally, the study reported here will have implications at a local level which
can translate to similar situations in local court systems. In the past, there has not been
any formalized way to sort future recidivists from non-recidivists when a youth comes
into contact with the court for committing a sex offense. The study reported here will
provide a process for county courts to use the J SOAP H to assess the risk for recidivism
that a youth poses. Risk score cutoffs will allow the court to sort youth into who pose

the least risk and can safely be treated in community-based programs and those who pose

35

the greatest risk and require more intensive treatment. This sorting frees resources that
would have been used in unnecessarily intensive treatment for the low risk youth so that
those resources can then be re—allocated to the high risk youth, possibly decreasing their

recidivism rates and increasing overall public safety.

Research Questions

This study will assess the predictive validity of the J SOAP H via the following

four research questions:

1) How well can the J SOAP H correctly identify a male juvenile who will reoffend

sexually (true-positive)?

2) How well does the J SOAP H differentiate between male J 805 who do recidivate

sexually and those who do not?

3) Does the J SOAP H have any utility in predicting non-sexual recidivism for male J S05?

4) Can the J SOAP 11 scores be divided into Low, Moderate, and High categories for male

J S03?

36

Methods

Sample

The sample uses secondary data drawn via collaboration between two medium
sized Midwestern Counties’ J SO treatment programs. All J SOAP H assessment results
were de-identiﬁed and provided to the research team after coding for both sites. The data
were analyzed anonymously. The ﬁnal sample consisted of 57 boys from the site 1 and
72 boys from the site 2, totaling 129 juveniles. The youth ﬁ'om site 1 were all youth
adjudicated of a sex offense in that county from 2002 through 2007. The youth from site
2 were all youth adjudicated for a sex offense in that county during 2005. The time
period for site 1 was selected to gain as large a sample as possible, allowing for 12 month
follow up after treatment completion. The time period for site 2 was also selected to
allow 12 month follow up after treatment completion. Treatment lengths varied across
youth and sites, varying from 6 to 24 months. Of the 129 juveniles, 60 of them were
charged with two offenses and 8 came in with three or more offenses. Their offenses
consisted of 59 charges of criminal sexual conduct in the ﬁrst degree, 46 charges of
criminal sexual conduct in the second degree charges, 16 charges of criminal sexual
conduct in the third degree, 15 charges of criminal sexual conduct in the fourth degree,
19 charges of indecent exposures, and 34 charges of gross indecency. The juveniles
range in age from 7 to 17, with an average of 13.7 and a standard deviation of 2.29.
There were originally135 males and 4 females (2 from each site) in this sample. The
females were excluded from analyses due to concerns of confounding gender differences
with different risk patterns, as previously mentioned. Additional demographics such as

race were not available in the data.

37.

The treatment pro grams of the two sites were analyzed to ensure that they were
relatively similar. Both programs claim to treat youth at their own pace, following a
“milestone of accomplishment model” rather than a “time to completion model”. Both
programs also use individual, family, and peer group treatment focusing on psychological
therapy and cognitive-behavioral restructuring. Both also claim to place additional
emphasis on sexual norms and boundaries and normative social skill development. The
treatment programs appeared to be similar enough to combine youth for treatment.
Analyses were additionally completed to be sure the male youth were not signiﬁcantly

different in any of the outcome variables.

Procedures and Training

Court workers at site 1 coded past cases since 2001 that were closed (completed
treatment) before January of 2008, allowing for at least a one year follow up period.
While they were aware of the treatment progress of the youth, they were told to base the
assessments only on what was known after their initial meetings with the youth. They

were blind to recidivism outcomes of the case.

Court workers were trained on using the J SOAP H and were instructed to code the
assessment based on the information available to them after their ﬁrst assessment
meetings, before treatment really began. The workers who coded the cases were the
actual workers assigned to the case by the court and had had several meetings with the
youth at the time of his/her entry into the court system for the sex offense. The workers

were instructed to only consider the youth’s state at the time of entry into the system, not

38

any progress made during treatment. They were also instructed not to code any case that

they did not remember or feel familiar enough with to give accurate responses.

The workers were trained using current cases to achieve an acceptable level of
inter-rater reliability. Training was done using taped interviews of actual cases. Court
workers listened to the tapes together and discussed how the J SOAP H should be coded
for that case. Training of this sort continued until individual workers were comfortable
using the instrument. Inter-rater reliability checks were then conducted, consisting of
court workers listening to taped interviews with youth and independently coding J SOAP
H risk assessments. This was an iterative process, repeating until workers were

comfortable with the instrument and inter-rater reliability reached .90.

At site 2, the sample was collected a little differently. Every program worker
completed an in-depth initial assessment which was developed over three to seven
meetings with a youth after he/she is initially sent to the treatment program. This initial
assessment is designed to aid in understanding the youth’s situation, plan the treatment
protocol, and present to the court. All cases were coded for youth who had their initial
assessment in the 2005 calendar year. This year was selected because it was the most
recent year that allowed for a one-year post treatment follow up (all youth who entered
the program in 2005 completed the program by March of 2008). Volunteers working
with this site were trained in using the measure and coded the cases based on the initial
assessment written report. This team of four volunteers coded the cases working in
rotating pairs, allowing them to consult with one another when they were unclear as to
how an item should be coded. Inter-rater reliability checks and trainings were also done

with the coding team (with each person coding cases separately) until a .90 level of

39

agreement was reached. This team coding set up provides increased conﬁdence in the

inter-rater reliability.

Recidivism was calculated for the one year (12 months) following the male
youth’s initial intake interview/process at each site. There were two females in each site
who were excluded, leaving a resulting sample size of 55 in the primary county and 70 in

the secondary count, totaling 125 combined.

This study intends to improve upon the existing literature in several ways. As
previously mentioned, there are very few studies on risk assessment measures for this
population. Thus, an additional study will add to the limited body of literature by testing
the validity of the J SOAP H for male J SOs. The primary critique of past studies
attempting to validate the J SOAP H is that they have samples that are unrepresentative of
how the measure is most useful. It is primarily a tool to assess the risk for recidivism of
male J 803. The primary population for this would be all male youth charged with a sex
offense at the county court level. Thus, studies that assess male J 805 who have
previously been “sorted” and are residents of a treatment facility (likely higher risk than
the general I SO population) or those who have been referred for sexual concerns but who
have not committed a sex offense, do not really assess the validity of using the measure

as it was intended.

This study assesses the predictive validity of the J SOAP H in the context that it
was intended for and would most likely be used for. Using the J SOAP H in a county
court setting complies with the standards set forth by the National Council on Crime and

Delinquency (1998) to test the measure for validity and reliability, while aiding in

40

making decision-making fair across male juveniles. The study reported here is also
extremely useful for the primary county in question, since they have requested using such
a measure to aid in predicting recidivism, deciding treatment modality, and
individualizing treatment plans, which was also recommended by the National Council

on Crime and Delinquency’s standards (1998).
Measures
J-SOAP II.

As previously detailed, the study reported here uses the J SO Assessment Protocol
II (J SOAP H). It is a 28-item measure intended to assess risk for recidivism for youth
who have been adjudicated for a sex offense. Each item is scored 0 for no presence, 1 for
some or suggested presence, or 2 for deﬁnite or extensive presence of the risk factor. The
measure consists of 4 subscales, Sexual Drive/Preoccupation, Impulsive/Antisocial
Behavior, Intervention, and Community Stability/Adjustment. The entirety of this

measure and its training manual is available without cost online.
Recidivism.

Recidivism data was collected by the individual counties and combined with the
de-identiﬁed J SOAP H. Both counties checked through their system and a youth’s new
crime was be recorded (actual crime type) if there was a new petition (charged offense) in
the 12 months following the youth’s initial crime. This is a possible limitation since only
new crimes committed within the same county were captured. This process appears to be
fairly standard across similar studies (Schwalbe, 2007). Recidivism data was coded
separately for sexual and nonsexual offenses, with 0 indicating no recidivism and l

41

indicating at least 1 new offense. In a larger related study of general delinquency in the
same county, a random selection of 10 cases were also checked against the Law
Enforcement Information Network (LEIN) system of national data and found that none of
the juveniles’ status as a recidivist or non-recidivist would change based on this wider

sampling, suggesting that checking for recidivism in the County system alone is adequate.

42

 

Results

There were 11 youth who were reported to have committed a sex offense and 22
youth who were reported to have committed a nonsexual offense during the follow-up
period. This resulted in a total recidivism rate of 23.2% for any type of offense, a sexual
recidivism rate of 8.8%, and a nonsexual recidivism rate of 17.6%. The two counties
were not signiﬁcantly different on any type of recidivism or total J SOAP score (as
calculated via an Independent Samples T-Test) and were thus analyzed as if they were
one population. For organizational purposes, the results will be presented by research

question.

1) How well can the J SOAP II correctly identify a male juvenile who will reoffend

sexually (true-positive)?

A Receiver Operating Characteristic Area Under the Curve (ROC AUC)
calculation was used to assess the ability of the measure to correctly classify male
juveniles who reoffend sexually. This resulted in a test statistic of .79 (p<.05) for total
I SOAP score, .76 (p<.05) for the Sexual Drive/Preoccupation subscale, .71 (p<.05) for
the Irnpulsive-Antisocial Behavior subscale, .63 (ns) for the Intervention subscale, and
.75 (p<.05) for the Community Stability subscale. This calculation corrects for base rate
by calculating the ratio of tr'ue-positives to false-positives, thus making it robust to the

low recidivism rate.

Correlations were also calculated between overall J SOAP score and recidivism as
well as for each subscale and recidivism. These are presented in Table 2. The total

J SOAP score and both static subscale scores were correlated with recidivism. Neither

43

dynamic subscale score (intervention or community stability) was signiﬁcantly correlated

with sexual recidivism.

2) How well does the J SOAP H differentiate between male J SOs who do recidivate

sexually and those who do not?

An Independent Samples T-test was used to look for differences in total J SOAP 11
score between those who recidivated sexually and those who did not. The groups were
signiﬁcantly different, t=-3.79, p<.05, dﬁl23. The average score for those who did not
recidivate sexually was 14.42 (sd=7.32) while the mean total score for sexual recidivists

was 23.3 (sd=8.l3).

3) Does the J SOAP H have any utility in predicting non-sexual recidivism for male J 805?

A Receiver Operating Characteristic Area Under the Curve calculation (ROC
AUC) was again done using I SOAP H total and subscale scores to predict non-sexual
recidivism. This resulted in a test statistic of .63 (ns) for total J SOAP score, .60 (ns) for
the Sexual Drive/Preoccupation subscale, .63 (p<.05) for the Irnpulsive-Antisocial
Behavior subscale, .53 (ns) for the Intervention subscale, and .64 (p<.05) for the
Community Stability subscale. This indicates that the J SOAP is not a good predictor of

non-sexual recidivism overall, but some scales appear to be better predictors than others.

Correlations were again calculated, looking for a relationship between overall
I SOAP H score, subscale scores and non-sexual recidivism. These rates are reported in

Table 3. No subscales were signiﬁcantly correlated with non-sexual recidivism.

44

4) Can the J SOAP II scores be divided into Low, Moderate, and High categories for male

J 8057

Youth were ﬁrst divided into the risk groups suggested by Prentky in his grant
report (2006), with low risk being scores under 25, moderate being scores 26 to 33 and a
high risk group for scores over 33. These results are presented in Table 4. This resulted
in 91% of the overall sample falling into the low risk category (with 9 youth in the
moderate category and 2 in the high risk category). An ROC AUC was done on this to
see how well these groups predict sexual recidivism. An area of .70 was found and was
signiﬁcant (p<.05). While this shows that the risk cut offs suggested do differentiate
youth based on risk, they lump most youth into the low risk category which may be less
helpful for considering differing levels of treatment. A One-way Analysis of Variance
(ANOVA) was calculated to see if these risk score cutoffs were signiﬁcantly different in
relationship to total, sexual, and non-sexual recidivism. All were signiﬁcantly different.

Results are presented in Table 5.

The distribution of scores and recidivism were then assessed to see if better cut
offs could be calculated. A histogram of the number of youth at each I SOAP score is
presented in Figure 1. This was then used alongside the ROC AUC calculation to ﬁnd
out offs that optimized the predictive ability of the measure for this population. The ideal
cut offs made low risk youth those with scores under 10, moderate risk 11 through 22,
and high 23 and above. Descriptives for these groups are presented in Table 6. This
resulted in a signiﬁcant ROC AUC of .80 (p=.05). A one-way Analysis of Variance
(ANOVA) was also done to see if the three risk groups were signiﬁcantly different based

on sexual recidivism. This resulted in an F-statistic of 9.8 with a p value of less than .05.

45

The between-groups variance, also known as the amount of variance in the dependent
variable explained by the independent variable, or omega squared, is .03 for low risk, .02
for moderate risk, and .01 for high risk. Given Cohen’s guidelines of .01 being a small
association (Cohen, 1994), .059 moderate association, and .138 a large association, all of
the risk groups in this sample have a low association. This means that a very small
portion of the total variance in sexual recidivism can be accounted for using these risk
groups. They were also signiﬁcantly different for predicting any recidivism with
F(1,126)=8.31 (p<.05). They were not signiﬁcantly different related to non-sexual

recidivism (F(1,126)=2.51 (ns)).

46

Discussion

This study adds to the existing literature suggesting the utility of the J SOAP II to
assess risk for sexual recidivism for male J 808. The signiﬁcantly high ROC AUCs for
sexual recidivism (.79 for total score) and high correlations (.32 for total score) are
another suggestion that the J SOAP H is a fairly good predictor of sexual recidivism. The
two static subscales appear to be better predictors than the two dynamic subscales, with a
nonsigrriﬁcant ROC AUC for the intervention subscale and sexual recidivism, although
the overall score appears to be the most predictive. Further analyses are needed to see
how this scale relates to sexual recidivism. The measure appears to be less effective in
predicting non-sexual recidivism, with a nonsigrriﬁcant ROC AUC (.63). This suggests
that a general delinquency risk assessment measure should still be used on this population
to predict non-sexual recidivism, especially given the non-sexual recidivism rate of
17.6%. The J SOAP also appears to do a good job of differentiating those male youth
who do recidivate sexually ﬁ'om those who do not. A 9 point total score difference was

found between those who recidivate sexually and those who do not.

The risk group cut offs presented by Prentky (2006) appear useful in
differentiating between sexual recidivists and non-recidivists, with a signiﬁcant ROC
AUC of .70, but there appear to be more appropriate and useful cutoffs for this
population since there are both prediction and treatment oriented goals. The risk group
cutoffs developed on this sample appear to better predict recidivism, with a signiﬁcant
ROC AUC of .80. This is similar the total score ROC AUC found in the literature

(Hecker et al., 2002; Martinez, Flores, & Rosenfeld, 2007; Prentky, 2006) of around .78

47

 

to .82 and signiﬁcantly better than some studies found (ROC AUC of .59, Viljoen et al.,

2008)

Given the literature suggesting some items that are most predictive of sexual
recidivism and should be included in risk assessments, the study reported here also
looked at the AUCs for each item for predicting sexual recidivism. This analysis showed
that 7 items were signiﬁcantly predictive of sexual recidivism: the number of victims,
duration of offense history, school behavior problems, having ever been charged or
arrested before the age of 16, understanding risk factors, management of sexual urges and
desires, and stability in school. These results are presented in Table 7. An AUC using a
total score from these 7 items to predict sexual recidivism was signiﬁcant (p<.05),
suggesting that these 7 items may be adequate predictors of sexual recidivism. The study
reported here suggests future research into this option. These items are reﬂective of some
of what was found in the literature, with Redlak (2003) ﬁnding signiﬁcant prediction
using items concerning number of victims and deviant sexuality, Miner (2002) suggesting
items regarding impulsivity (possibly measuring the same construct as school behavior
problems or stability in school), and higher numbers of prior sex offenses. Items
suggested to be predictive by Redlak (2003) that were not found here were: history of
sexual abuse, adverse family climate, and out-of-home placement. Miner (2002) also
suggested some items that were not reﬂected in this study, namely: had a male victim,
and present paraphilias and Clyburn’s study (2002) found items that were not at all
represented in The study reported here of the J SOAP H: age, 2+ clinical diagnoses,
altercations (both physical and verbal) while in treatment, and length of treatment. The

study reported here had several items in common with the studies found in the literature,

48

but more research is needed to assess what factors are most related to criminogenic risk

for juvenile sex offenders.

This study adds to and improves upon the existing literature in several ways. As
previously noted, the 14 known studies on J SO risk assessments all used unrepresentative
samples with varying follow-up lengths. The study reported here improves upon this
standard by using a representative sample of all sex offending boys coming into contact
with two county courts. It also uses a stable follow-up length, avoiding differential time

at risk.

With the results of this study, as similar to many studies on this population, more
questions and calls for future research are brought about. While the study reported here
improves upon many of the shortcomings noted of existing literature, more work is still
needed to test this and other J SO risk assessment measures prospectively with
representative populations. The next step of the study reported here is to fully implement
the use of the J SOAP H in the assessment and decision making process of one of the
counties included in this study. This will allow for prospective analysis of the validity of
the measure, without concerns of coders knowing any treatment outcomes. More work is
also needed to assess what other aspects of J SO’s lives are related to crirrrinogenic risk.

It is hoped that studying the contextual community factors in the areas where these boys
spend their times will allow better understanding of the population and more
comprehensive analysis of risk. Geospatial mapping technology could be useful in
assessing these contextual factors and possibly combining them with risk assessment

variables to form a multivariate analysis of criminogenic risk for J 803.

49

Additional research is also needed to better understand ﬁrture sex offending
behaviors of these youth. Not all crime is reported to police and captured in ofﬁcial
reporting data. In order to have a complete picture of the likelihood of reoffense outside
sources such as self-report and family/neighbor/peer report should be utilized. Without
understanding what is not captured in ofﬁcial reports and why there may be differential

reporting/prosecuting, true risk for recidivism cannot be fully assessed.

Similarly, on a greater scale, our knowledge of adolescent sexual behavior is still
limited. It is unknown how much sexual offending behavior is truly happening in
society. It is understood that the boys who are represented in the adjudicated court
system, even as ﬁrst time offenders, are not the full population of those engaging in
illegal sexual behavior. More research is needed to assess what is normative sexual
behavior for adolescent boys and how much illegal sexual behavior is happening outside
of the adjudicated p0pulation. With this, more sex offense prevention programs aimed at
adolescent male perpetrators are needed to reduce the number of offenses that are

occurring.

As noted in the review of the literature, this is an understudied population across
variables. More research is also needed to better understand this population as related to
criminogenic risk. More research is needed to address treatment efﬁcacy concerns. A
review of the treatment studies on juvenile sex offenders is not currently available in the
literature and would be greatly beneﬁcial. Since this population is majority male, more
research on the females that are present is needed and an improved understanding of the
dynamics of gender and sexual behaviors in general is also called for. More is needed to
be known about this population as a whole.

50

Across sub-topics within the juvenile sex offender literature, improved rigor and
quality is needed. As previously detailed, there are relatively few published studies on
juvenile sex offender topics. There are quite a few theses and dissertations, but there
appears to be a disconnect between such studies and published journal articles. Such may
understandable given the limitations that must be noted with this research, as will be
discussed below. More research is needed to be made available within the existing

published literature on juvenile sex offender topics.

As mentioned, any research on this speciﬁc population comes with several
limitations, which may make publication difﬁcult. Sex offenses are a relatively rare
event in the court. While there may be frequent juvenile sex offenses in the public, there
is a sorting that occurs resulting in only a portion coming into contact with the court. It is
unknown if this sorting is random or if there are speciﬁc factors (which may or may not
be related to risk for recidivism) that result in differential reporting and prosecuting of

offenders.

Site differences may also limit the interpretability of this study. Sites were
selected based on their description of pro grammatic components. It is unclear if any
differences were present. Treatment modalities and dosage/intensities may differ slightly
between the two sites and also between youth. Since treatment length and modality
differ, actual time at risk may differ as well. This was not accounted for in the study

reported here and may affect recidivism.

Due to the desire to keep the sample size as large as possible, boys with different

types of sex crimes were lumped together for analyses. It is possible that youth with

51

different crime types differ greatly in regards to recidivism. This is difﬁcult to study due
to small sample sizes and also due to possible and unknown difference between the actual
type of crime committed and the crime that the youth was eventually charged with. Just
as there are factors that affect who actually comes into contact with the court, there are
factors that affect what crime youth are charged with. The crime type that is represented

in the data may not be reﬂective of the actual acts that were committed.

There is also a low base rate of recidivism, making statistical inference difﬁcult.
Ofﬁcial reports were used to calculate recidivism in this study. It is unknown how much
of an underestimate of the true recidivism rate this actually is. The generalizability of
results is also difﬁcult since the study reported here is still of relatively small scale and
short follow-up. Larger samples and longer follow-up periods, controlling for disposition

and time at risk are needed.

The utility of this study is also of concern. While results are statistically
signiﬁcant, they may not be of practical usefulness. It is unclear if using the J SOAP H
adds any incremental validity, and thus has any added usefulness, above and beyond
general risk assessment measures. The risk score cutoffs also have cautionary utility clue
to their lack of cross validation. While it is supported through the study reported here to
say that low risk males on the J SOAP H are very unlikely to recidivate, practically, it is
still much safer consider all J $05 as high risk. There is no acceptable level of false-
negatives, or youth who are labeled low risk and then recidivate. There is also the fear of
the unknown recidivism. Since sexual recidivism rates are likely underestimates of true
occurrences, risk levels based on ofﬁcial recidivism may not be reﬂective of who is truly

risky. Therefore the actual utility of such risk levels is questionable.

52

However, even with these limitations, the study reported here shows the
promising utility of the J SOAP H in this population of male J 803. This research is a step
in the needed direction of more research on this population. Further research is needed
prospectively, on greater scales, with larger sample sizes and greater location areas.
Research is also needed to assess gender differences since there is a small proportion of
J 808 who are female. Future research should also include contextual aspects including
family and community variables. In order to fully understand a J SO’s risk for recidivism,
wide studies of the general population including inhibitory factors for reporting and

prosecuting will be necessary.

53

Table 1. Juvenile Sex Offender Risk Assessment Literature Summary

 

 

 

 

 

w
r ) w .m m
m m e um m mm
2 g V. I .I
m on w m. M m n .m m m
w m N m m... s .m m .Ia m .Is
a m m e pr. .m a m. m m m m w
e u a 8 v a o e e a O a e
M A S A A M L L S R T R R
.c<m:..m Ewan—2. mm m- HPN 22 :52 02 . 3 Ex. 3.3» 8 oa um .wm 3.3-9.63. 8212.0:
mmx 1219 No 3:313? .92 303?». .w e.6 um 3. 34 5.83 amazon: o<m3= «83 mi.
02253 3.2m... m. 398833..” 2:03 8n.n..<.m3 mi 80233. Dmmnqﬁnzmm
amx x533- v.83? <05: Enigma amp... 53 o.“ 308 aimqmanmm 322mm:
Emma- :a 3003 3.622. m 38 2mm m<m..mc.m. 335.33. 2:. no: Enigma
32: 233.2: 838: m<m..mc_m. o:.<
3088. NE 3.593%...
Cmo>p a {w :03 $5.38
$3.3.
$3.3. wa z 2.» 2.... =<m2 353-2% 0 <3; P93 SH 0* munqox.3m8.< Ii: mnm.m N <05: imam w
23.8.40? w .c<m:..m max Eu. 8* ".32. 304m 58... 3
5.9.8. ommaamqm: «95.53 $53.? 3m: .92
3%.. mnm.m N <23
gnaw?
<m< m.
992:
$003
Imnxm: ma 2 2m x 5.3323 32m 8-5 5*. Am o.ﬂ 3X. .No 0." Sm moam H >cn .3. 83. mnoqm
mnocE: r max ommaama <33. 9m m3 m3 so” 843.33 2.3 mmxcm.
2mg..- «mnazaa
man m,
2m:m_m
308.

 

 

 

 

 

 

 

 

 

 

 

 

S4

Table 1 cont

 

 

 

 

 

 

 

 

7:553. mo 5 HP... x 2.0.0.803. mmx 23 mex .m 0." 3.x. .5 0. mo. .3 820.0203 002,503 38.
2030. an - 0.403004 mo. mnoqm 033 mmxcm. 28.02.03.
W825... 3 £033.03" .um >cn .0. 83.. .808 030
325 0.333. 052 3.20. «2.02.03... .8 >cn
mow» .0330. 3x. .04 3.303..” .803 030 8.80.
>383 >303nm3. 28.0253
me nmcnmm.m3
..:<m3..m wen—G mu Hmm 5 Sam x mmncqm 22 mbvm web...» .90 50.9.0 23 03.0 3
mmx mm:— - 3:02.030. 0.2033308 0032003 203
0.403qu Soon: 3 31.3 E3 3 .30 0. 0433003 030 30
2% 0.303030 02.80 .30:.m.<m\>3n.mon.m.
>303- 0.d .c<m3..m 33.3 3033.2 2.3.30.0 20m
2.03" 0.00.2.5 0.. 0980.
.0333. .. 323.323 Anox 393.563 I
awoke .c was om .wo. 060$.
.<.nno< 5m 2 22 2.. 0:83.03. HTS Wax max. moam 330.080 238.0...
3003 3 3.03303 8334 <33 3.: 30" m.m3.mnm3n.<
033.220
3.0322 you w- 5.3 <05... .3 Sm m EX :3 3.3.. $3 04 >cn om .33 40..
.Noom.l No 9.303 0. Sm 3033a- 04 um: um: 0.030.383; 033 .mow 8.
30352.- 00003303" 0.ﬂ w <32... 000.8838
0” 000.0. $2.an .3
03003an memmnzcmmzm
mm m «33" £30 (<03
.6008. 883303000 3
30" <0" .00 05.5an 40..
05.74.00 0920.... 0.0520
.3 0 0032.03
0333.03
0. .0530.
.633 2.» 2 22 22 22 z» z» 2.. m.m3.3nm3. 3.2.0336
ﬂood 3 0032003 3.3033 303 030
$30. «mn.3.<.03
5:00? How 5 23 x 203.32% m.mm max. 3.3» .mm >cn 34 «980. Sn.3.<.m3
23.08. .w 3.0.0035. $3.3M <00;

 

 

 

 

 

 

 

 

 

 

 

55

Table 1 cont

 

 

 

 

 

 

 

 

 

 

 

9008. 0 .3 23020303

0000... 300.03 300 02

9:22.

C=:63.w

002838

z .Noom.
Znn0< 50 z 2» 23 0080303" Pmﬁw 90.x. mmx m030 820.080 00008.00
3000. 3 200303" 83.2 <00; 00. :0. 053.383?
080.220
350» 10303 S 2 22 2» 30000330. 20 z» 2» 0.2033380 0022003
Good 3 202302 830“ 00x00. 8.0.0.303 030 303-
. 8202.29.

202.5 5m 2 2.» x 0m 93303.2- 22 z.» z» >cn 0. .310“ 00x00.

Soon. 0 00000 0m03n.0m ..0n.0.<.v.3 00.3« 88. 033.
030 mp «00.00330. .mm >cn moq n..3.8. 3%
202303" 8303 003308
.030<00:.0n
00n3_0n0303w.

._-mOmw>... €002.63 mum z z.» x 00.00.9080 30.0... 2.» 20 Zn >cn 0.“ .mm .9 00x00.
0. 0.. m Sn.0.<.m3 0m 0 .0<03._0 030
$008 .00 3.. 00x00. «0902.03 00
03000:

5:003. 50 pm 23 x 203-0028 9mm max 30.x. .mm >cn 8.. 00x00. «0902.03

m8.08. .w “00.00330. $2.3m <00;

0000.0. 0 .3 73.020033

m000«~ 300.03 300 92 .

n30<0~.

czqa3.

$8,503.”

0380.

 

 

 

 

 

 

 

 

.Zmuzoﬁ 3000300

56

Table 2. Correlations Between J SOAP and Sexual Recidivism.

 

 

 

 

 

 

 

Total Score Sexual Drive/ Impulsive- Intervention Community
Preoccupation Antisocial Subscale Stability
Subscale Behavior Subscale
Subscale
Sexual .32 (p<.05) .27 (p<.05) .25 (p<.05) .14 (ns) .13 (ns)
Recidivism

 

 

S7

 

 

Table 3. Correlations Between J SOAP and Non-Sexual Recidivism.

 

 

 

 

 

 

 

lTotal Sexual Drive/ .Impulsive- Intervention Community
Score Preoccupation Antisocial Subscale Stability
Subscale Behavior Subscale
Subscale
Non-Sexual .16 (ns) .10 (ns) .13 (ns) .06 (ns) -.06 (ns)
Recidivism

 

 

58

 

Table 4. Prentky Inspired Risk Groups

 

 

 

 

 

 

 

 

 

Group N Total J SOAP Total Sexual Non-Sexual
Score Recidivism Recidivism Recidivism

Low 114 (91.2%) <25 19% 5% 15%

Moderate 9 (7.2%) 26-33 67% 44% 53%

High 2 (1.6%) 34+ 50% 50% 0%

 

 

59

 

Table 5. AN OVA for Prentky Inspired Risk Groups

 

 

 

 

 

F -Statistic Signiﬁcance
Any Recidivism 6.08 p<.05
Sexual Recidivism 11.80 p<.05
Non-Sexual 5.27 p<.05
Recidivism

 

 

 

60

 

Figure 1. Number of Youth at Each J SOAP Score

Number of Youth at Each JSOAP Score

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

15.. Mean=1S.36
St.Dev=7.72
”"1
r...
1o~ {a --1 . r—
>. -
g m . .Iq
a, "'1
3
8' r-—- ~—
L .
LL
5—
0 ﬂ ., i i l K ’ 1 l ' l
o 5 1o 15 20 25 4o

 

Overall JSOAP Score

61

Table 6. Empirical Risk Group Descriptives

 

 

 

 

 

Group N Mean Total Total Sexual Non-Sexual
J SOAP Score Recidivism Recidivism Recidivism
Low 32 5.81 12.5% 0% 12.5%
(25.6%)
Moderate 70 (56%) 15.66 18.6% 5.7% 14.3%
High 23 26.87 52.2% 30.4% 34.8%
(1 8.5%)

 

 

 

 

 

 

62

 

Table 7. Signiﬁcant Items for Predicting Sexual Recidivism

 

Item
l-Prior Sex Offense
Z-Number of Victims
3-Male Child Victim
4-Duration of Offense History
5-Degree of Planning
6-Sexualized Aggression
7-Sexual Drive/Preoccupation
8-Sexual Victimization History
9-Caregiver Consistency
lO-Pervasive Anger
ll-School Behavior Problems
12-Hx of Conduct Disorder
l3-Juvenile Antisocial Behavior
14-Ever Charged/Arrested before age 16
lS-Multiple Types of Offenses
16—History of Physical Assault/Family Violence
17-Accepting Responsibility for Offense
18-Internal Motivation for Change
19-Understands Risk Factors
20-Empathy

21-Remorse and Guilt

 

 

AUC

.57

.73

.46

.71

.58

.51

.66

.61

.58

.62

.71

.62

.53

.70

.59

.52

.55

.59

.67

.62

.65

 

Signiﬁcance
ns
p<.05
ns

p<.05

"S

"S
"S

p<.05

p<.05
ns
ns
ns
ns

p<.05
ns

ns

 

63

 

Table 7 Continued

 

 

Item

22-Cognitive Distortions

23-Quality of Peer Relationships
24-Management of Sexual Urges and Desire
25-Management of Anger

26-Stability of Current Living Situation
27-Stability in School

28-Evidence of Positive Support Systems

 

AUC

.60

.49

.71

.65

.65

.69

.59

Signiﬁcance
ns

ns

p<.05

HS

p<.05

 

 

64

 

References

Anaforian, C. (2009). Predicting treatment completion of adolescent sex offenders in a
community outpatient setting. Alliant International University.

Andrade, J. T., Vincent, G. M., & Saleh, F. M. (2006). Juvenile sex offenders: A complex
population. Journal of Forensic Sciences, 51(1), 163-167.

Barbaree, H. & Marshall, W. (2008). The Juvenile Sex Oﬂender. The Guilford Press:
New York, NY.

Becker, J., Harris, C., Sales, B. (1993) Juveniles who commit sexual offenses: A critical
review of research in Sexual Aggression by Hall (Ed).

Blaskeet, D. (1989) In Barabee & Marshall (2008). The Juvenile Sex Offender. The
Guilford Press: New York, NY.

Bonta, J. (1996). Risk-needs assessment and treatment. In A.T. Harland (Bd.), Choosing
correctional options that work (pp. 18-32). Thousand Oaks, CA: Sage.

Borum, R. & Verhaagen, D. (2006). Assessing and managing violence risk in juveniles.
The Guilford Press: New York, NY.

Bourke, M. L., & Donahue, B. (1996). Assessment and treatment of Juvenile Sex
Offenders: An empirical review. Journal of Child Sexual Abuse, 5(1), 47-71.

Burke, D. (2001). Empathy in sexually offending and nonoffending adolescent males.
Journal of Interpersonal Violence. 16(3), pp. 222.

Burton, D., Smith-Darden, J ., and Frankel, S. (2006). Research on adolescent sexual
abuser treatment programs. In Barabee and Marshall (2008).

65

Burton, D. (2000). Were adolescent sexual offenders children with sexual behavior
problems?. Sexual Abuse: A Journal of Research and Treatment, 12(1), 37-48.

Buttler, S & Seto, M. (2002). Distinguishing two types of adolescent sex offenders.
Journal of the American Academy of Child & Adolescent Psychiatry. 41(1), pp.
83-90.

Caldwell, M. (2002). What we do not know about juvenile sexual reoffense risk. Child
Maltreatment. 7(4), 294.

Calley, N. G. (2007). Promoting an outcomes-based treatment milieu for juvenile sexual
offenders: A guided approach to assessment. Journal of Mental Health
Counseling, 29(2), 121-144.

Casines, L. (2003). A transitional day treatment programmeme for juvenile sexual
offenders. Carlos Arlos Albizu University.

Center for Sex Offender Management. (2007). The comprehensive assessment protocol:
A systemwide review of adult and Juvenile Sex Offender management strategies.
United States Department of Justice, Ofﬁce of Justice Programs. Bureau of
Justice Assistance.

Clyburn, T. (2002) A three-year study of multivariate predictors and outcome measures
for adolescent sex offenders. Dissertation Abstracts International. 62(9-B).

Cohen, J. (1994). The earth is round (p<.05). American Psychologist, 49, pp. 997-1003.

Cooper, P. (1997). Residential treatment programmeme criteria affecting the length of
stay of adolescent sex offenders. Dissertation Abstracts International. Vol 58(4-
A).

66

Curwen, T. (2003). The importance of offense characteristics, victimization history,
hostility, and social desirability in assessing empathy of male adolescent sex
offenders. Sexual Abuse. 15(4), pp. 347.

Daversa, M. (2005). Early caregiver instability and maltreatment experiences in the
prediction of age of victims of adolescent sexual offenders. Fielding Graduate
Institute.

Davis, G. and Leitenberg, H. (1987). Adolescent sex offenders. Psychological Bulletin.
101(3), pp. 417-427. '

DeNatale, R. (1989). An investigation of demographic, emotional, and attitudinal
indicators of male Juvenile Sex Offenders. Unpublished thesis. Dissertation
Abstracts International.

Durham, A. (2006). Young men who have sexually abused. West Sussex, England: John
Wiley & Sons Ltd.

Eastman, B. J. (2005). Variables associated with treatment failure among adolescent sex
offenders. Journal of Oﬁ'ender Rehabilitation, 42(3), 23-40.

Eﬁa—Breitbach, J. and Freeman, K. (2004). Treatment of juveniles who sexually offend:
An overview. Journal of Child Sexual Abuse. 13(3/4), pp. 125-138.

Epperson, D., Ralston, C., Fowers, D., DeWitt, J ., and Gore, K. (2005). Actuarial risk
assessment with juveniles who offend sexually: Development of the juvenile
sexual offense recidivism risk assessment tool-II (J -SORRAT-II). In D. Prescott
(Ed.) Risk assessment: Theory, controversy, and emerging strategies of youth who
have sexually abused (pp. 118-169). Oklahoma City, OK: Wood and Barnes
Publishing.

Epps, K. (1994). Treating adolescent sex offenders in secure conditions: The experience
at Glenthome Centre. Journal of Adolescence, 1 7, 105—122.

67

Erooga, M. and Masson, H. (2006). Children and young people who sexually abuse
others (2“d ed). Abingdon, ox: Routledge.

Federal Bureau of Investigation. (2000). Universal Crime Report Summary.

F inkelhor, D. (1984). Child Sexual Abuse: New Theory and Research. New York: Free
Press.

Flores, A. W., Travis, L. F., Latessa, E. J. (2003). Case classiﬁcation for juvenile -
corrections: An assessment of the youth level of service/case management
inventory. Cincinnati, OH: Division of Criminal Justice.

Ford and Linney. (1995). In Martin and Kline Pruett (1998). Juvenile sex offenders and
the juvenile justice system. American Criminal Law Review. 35(2), pp. 279-332.

Friedrich, W., Gerber, P., Koplin, B., Davis, J ., Giese, J ., Mykelbust, C., & Franckowiak,
D. (2001). Multimodal assessment of dissociation in adolescents: Inpatients and
juvenile sex offenders. Sexual Abuse. 13(3), pp. 167.

Funk, S. (1999). Risk assessment for juveniles on probation: A focus on gender.
Criminal Justice and Behavior, 26, 44-68.

Gibson, C. and Vandiver, D. (2008). Juvenile Sex Oﬂenders: What the public needs to
know. Westport, CT: Greenwood Publishing Group.

Grisso, T. (1996). Society’s retributive response to juvenile violence: A developmental
perspective. Law and Human Behavior. 20, pp. 229-247.

Hecker, J ., Scoular, J ., Righthand, S., and Nangle, D. (2002). Predictive validity of the J-
SOAP over I 0-plus years: Implications for risk assessment. Paper presented at the
Annual Meeting of the Association for the Treatment of Sexual Abusers,
Montreal, Quebec, Canada.

68

Heilbrun, K. (1997). Prediction versus management models relevant to risk assessment:
the importance of legal decision-making context. Law and Human Behavior,
21(4), 347-359.

Herkov, M., Gynther, Thomas, S., & Myers, W. (1996). MMPI differences among
adolescent impatients, rapists, sodornists, and sexual abusers. Journal of
Personality Assessment. 66(1), pp. 81.

Hersant, J. (2007). Risk assessment of juvenile sex offender reoffense. Southern Illinois
University at Carbondale.

Hoge, D. and Andrews, A. (1996). An investigation of risk and protective factors in a
sample of youthful offenders. Journal of Child Psychology and Psychiatry. 37(4),
pp. 41 9-424.

Hunter, J ., Figueredo, A., Malamuth, N., and Becker, J. (2003). Juvenile Sex Offenders:
Toward the development of a typology. Sexual Abuse: A Journal of Research and
Treatment, 15(1), 27-48.

Institute for Social Research. (2003). In Barabee and Marshall (2008). The Juvenile Sex
Oﬂender. The Guilford Press: New York, NY.

Kraemer, 3., Salisbury, S., & Spielrnan, C. (1998). Pretreatment variables associated with
treatment failure in a residential juvenile sex-offender program. Criminal Justice
and Behavior. 25(2), pp. 190.

Kubic, E., Hecker, J ., & Righthand, S. (2002). Adolescent females who have sexually
offended: A comparison with adolescent males who have sexually offended.
Journal of Child Sexual Abuse. 11(3), pp. 63-68.

LaFond, J. (2005). Preventing sexual violence: How society should cope with sex
oﬂenders. Washington, DC: American Psychological Association.

69

 

Langstrom, N., & Grann, M. (2000). Risk for criminal recidivism among young sex
offenders. Journal of Interpersonal Violence, 15(8), 855-871.

Leguizamo, A. (2000). Juvenile sex offenders: An object relations approach. Dissertation
Abstracts International 613-B.

Levenson, J ., Brannon, Y., F ortney, T., & Baker, J. (2007). Public perceptions about
Juvenile Sex Offenders and community protection policies. Analyses of Social
Issues and Public Policy, 7(1), 1-25.

Lindmeier, (2002). In Barbaree, H. & Marshall, W. (2008). The Juvenile Sex Oﬂender.
The Guilford Press: New York, NY.

Lindsey, R., Carlozzi, A., & Bells, G. (2001). Differences in the dispositional empathy of
juvenile sex offenders, non-sex-offending delinquent juveniles, and nondelinquent
juveniles. Journal of Interpersonal Violence, 1 6, 510—522.

Losel, F. & Schmucker, M. (2005). The effectiveness of treatment for sexual offenders: A
comprehensive meta-analysis. Journal of Experimental Criminology. 1, pp. 117-
146.

Lowenstein, L. (2006). Aspects of young sex abusers - A review of literature concerning
young sex abusers (1996-2004). Clinical Psychology and Psychotherapy, I3, 47-
55.

Martinez, R., Flores, J ., Rosenfeld, B. (2007). Validity of the juvenile sex offender
assessment protocol 11 (J -SOAP-II) in a sample of urban minority youth. Criminal
Justice and Behavior, 34(10), 1284-1295.

Maring, J. (1998). Using the Million Adolescent Personality Inventory to differentiate
sexually offending from nonsexually offending conduct-disordered adolescents.
Dissertation Abstracts International. 5 9(2-B).

70

Martin E. and Kline Pruett, M. (Winter 1998). Juvenile sex offenders and the juvenile
justice system. American Criminal Law Review. 35(2), pp. 279-332.

McCoy, W. (2008). Predicting treatment outcome and recidivism among juvenile sex
offenders: The utility of the J SOAP-II and ERASOR in an outpatient treatment
program. Unpublished doctoral dissertation, Sam Houston State University.

McIntyre, B. (2009). Do prior experiences and psychological states contributing to
juvenile sexual offenses differentiate between African American and Caucasian
American offenders?. The University of Memphis. Dissertation Abstracts
International. 7 0(3-B).

McMackin, R., Leisen, M., Cusack, J ., LaFratta, J ., & Litwin, P. (2002). The relationship
of trauma exposure to sex offending behavior among male juvenile offenders.
Journal of Child Sexual Abuse. 11(2), pp. 25.

Michigan Penal Code. (2009). <www.legislature.mi.gov>.

Miner, M. (2002). Factors associated with recidivism in juveniles: an analysis of serious
juvenile sex offenders, Journal of Research in Crime and Delinquency, 39, pp.
421-436.

Miner, M. and Munns, R. (2005). Isolation and normlessness: Attitudinal comparisons of
adolescent sex offenders, juvenile offenders, and nondelinquents. International
Journal of Oﬂender Therapy and Comparative Criminology. 49(5), pp. 491-504.

Mofﬁtt, T. (1993). Adolescence-limited and life-course-persistent antisocial behavior: A
developmental taxonomy. Psychological Review, 100, 674-701.

Monto, M., Zgourides, G., Harris, D. & Wilson, J. (1994). Empathy and adolescent male
sexual offenders. Perceptual and Motor Skills. 79, pp. 1598.

71

 

Nagayama Hall, G. (1995). Sexual offender recidivism revisited: A meta-analysis of
recent recidivism treatment studies. Journal of Consulting and Clinical
Psychology. 63(5), pp. 802-809.

National Council on Crime and Delinquency. (1998). Risk Assessment Standards.

National Task Force on Juvenile Sex Offending. (1993). Juvenile and Family Court
Journal. v. 55.

Ofﬁce of Juvenile Justice and Delinquency Prevention. (2009). Retrieved from
www.ojjdp.ncjrs. gov.

Onifade, B, Davidson, W., Campbell, C., Turke, G., Malinowski, J ., and Turner, K.
(2008). Predicting recidivism in probationers with the youth level of service case
management inventory (YLS/CMI). Criminal Justice and Behavior. 35(4), pp.
474-483.

Onifade, B, Davidson, W., Livsey, S., Turke, G., Horton, C., Malinowski, J ., Atkinson,
D., and Wimberly, D. (2006). Risk assessment: Indentifying patterns of risk in
young offenders with the youth level of service/case management inventory.
Journal of Criminal Justice. 36(2), p. 165.

Page, C., Tourigny, M., Renaud, P. (2009). A comparative analysis of youth sex
offenders and non-offender peers: Is there a difference in their coping strategies?.
Sexologies.

Parks, G. & Bard, D. (2006). Risk factors for adolescent sex offender recidivism:
Evaluation of predictive factors and comparison of three groups based upon
victim type. Sexual Abuse: A Journal of Research and Treatment. 18(4), pp. 319-
342.

Perry, G. & Orchard, J. (1992). Assessment and treatment of adolescent sex offenders.
Professional Resource Press, Sarasota, FL.

72

Petersen, J., Davidson, W., Campbell, C., Onifade, E. (2010). Exploring the predictive
validity of juvenile sex offender risk assessment instruments: A meta-analysis.
Manuscript in progress.

Pithers, W., Gray, A., Busconi, A., and Houchens, P. (1998). Caregivers of children with
sexual behavior problems: Psychological and familial functioning. Child Abuse &
Neglect, 22(2), 129-141.

Prentky, R. (2006). Risk management of sexually abusive youth: A follow up study.
Unpublished grant report for the National Institute of Justice.

Prentky, R., Harris, B., Frizzell, K., and Righthand, S. (2000). An actuarial procedure for
assessing risk with juvenile sex offenders. Sexual Abuse: A Journal of Research
and Treatment, 12(2): 71-93.

Prentky, R. and Righthand, S. (2003). Juvenile sex offender assessment protocol-II
manual. NCJ 202316.

Redlak, A. (2003). An exploratory meta-analysis of the predictor variables of juvenile sex
offenders who sexually recidivate. Unpublished doctoral dissertation. California
School of Professional Psychology.

Reitzel, L. & Carbonell, J. (2006). The effectiveness of sexual offender treatment for
juveniles as measured by recidivism: A meta-analysis. Sex Abuse. 18, pp. 401-
421.

Righthand, S., Prentky, R., Knight, R., Carpenter, E., Hecker, J ., and Nangle, D. (2005).
Factor structure and validation of the juvenile sex offender assessment protocol
(J -SOAP). Sexual Abuse: A Journal of Research and Treatment. 17(1), pp. 13-30.

Righthand, S., & Welch, C. (2001). Juveniles who have sexually offended: A review of
the professional literature. Oﬁice of Juvenile Justice and Delinquency Prevention,
1 -86.

73

Sapp, A. and Vaughn, M. (1990). Juvenile sex offender treatment at state-operated
correctional institutions. International Journal of Offender Therapy and
Comparative Criminology. 34, pp. 131-146.

Seagrave, D., & Grisso, T. (2002). Adolescent development and the measurement of
juvenile psychopathy. Law and Human Behavior, 26(2), 219-239.

Schmidt, F., Hoge, R. and Gomes, L. (2005). Reliability and validity analyses of the
youth level of service/case management inventory. Criminal Justice and
Behavior. 32(3), pp. 329-344.

Schwalbe, C. S. (2007). Risk assessment for juvenile justice: A meta-analysis. Law
Human Behavior, 31, 449-462.

Shaw, C. and McKay, H. (1969). Juvenile delinquency and urban areas (3rd ed.).
Chicago, IL: The University of Chicago Press.

Skaggs, R. (1997). Music-centered creative arts in a sex offender treatment programme
for male juveniles. Music Therapy Perspectives, 15(2), pp. 73-78.

Snyder, H. (2002). Juvenile Arrests 2000. Washington, DC: Ofﬁce of Juvenile Justice
and Delinquency Prevention.

Stevenson, M., Sorenson, K., Smith, A., Sekely, A., & Dzwairo, R. (2009). Effects of
defendant and victim race on perceptions of juvenile sex offenders. Behavioral
Sciences & the Law. 27(6), pp. 957-979.

Terry, K. (2006). Sexual oﬂenses and offenders: Theory, practice, and policy. Belmont,
CA: Wadsworth.

Tomatis, A. (2006). Using the BASC parent rating and self-report composite measures to
differentiate juvenile sex offenders from delinquent and non-sex offending
adolescents. Dissertation Abstracts International. 6 7(1 O-A).

74

Turk, A. (1969). Criminality and the Legal Order. Chicago: Rand-McNally.

United States Department of Justice SMART Ofﬁce. (2007). Retrieved from
www.ojp.usdoj.gov/smart.

United States Census Bureau. (2000). Retrieved ﬁom www.census. gov.

Vaih-Koch, S., Ponseti, J ., Bonsinkski, H. (2001). ADHD und storung des
sozialverhaltens im kindesalter als pradiktoren aggressiver sexualdelinquenz?
Sexuologie, 8(1), 1-18.

Vandiver, D. (2006). A prospective analysis of juvenile male sex offenders. Journal of
Interpersonal Violence. 21(5), pp. 673-688.

van Wijk, A., van Horn, J ., Bullens, R., Bijleveld, C., & Doreleij ers, T. (2005). Juvenile
sex offenders: A group of their own? International Journal of Oﬂender Therapy
and Comparative Criminology, 49(1), 25-36.

Viljoen, J ., Scalora, M., Cuadra, L., Bader, S., Chavez, V., Ullrnan, D., and Lawrence, L.
(2008). Assessing risk for violence in adolescents who have sexually offended: A
comparison of the J -SOAP-II, J -SORRAT-II, and SAVRY. Criminal Justice and
Behavior, 35(1), 5-23.

Vitacco, M. & Rogers, R. (2009) The assessment of psychopathy and response styles in
sex offenders, in Sex Oﬂenders: Identiﬁcation, Risk Assessment, Treatment and
Legal Issues. Saleh, F., Grudzinskas, A., Bradford, J., & Brodsky, D. (Eds).
Oxford University Press: New York.

Vold, G. (1958). Theoretical Criminology. New York: Oxford University Press.

Waite, D., Pinkerton, R., Wieckowski, E., McGarvey, E., and Brown, G. (2002).
Tracking treatment outcome among juvenile sexual oﬂenders: A nine year follow-

75

 

up study. Paper presented at the Annual Meeting of the Association for the
Treatment of Sexual Abusers, Montreal, Quebec, Canada.

Wieckowski, E., Hartsoe, P., Mayer, A. & Shortz, J. (1998). Deviant sexual behaviour in
children and young adolescents: frequency and patterns. Sexual Abuse: A Journal
of Research and Treatment, 10(4), 293-303.

Woodhams, J ., Gillett, R., & Grant, T. (2007). Understanding the factors that affect the
severity of juvenile stranger sex offenses: The effect of victim characteristics and
number of suspects. Journal of Interpersonal Violence, 22(2), 218-237.

Woods, J. (1997). Breaking the cycle of abuse and abusing: Individual psychotherapy for
juvenile sex offenders. Clinical Child Psychology and Psychiatry. 2(3), pp. 379.

Worling, J ., Curwen, T. (2004). The estimate of risk of adolescent sexual offense
recidivism (ERASOR): Preliminary psychometric data. Sexual Abuse: A Journal
of Research and Treatment, 16, 235-254.

Worling, J ., Curwen, T. (2001). The ERASOR: Estimate of risk of adolescent sexual
oﬂense recidivism (version 2.0). Toronto, Ontario, Canada: Safe-T Program,
Thisletown Regional Centre.

Young, S. (2007) Psychopathy as a predictor of recidivism among adolescent male sexual
offenders treated in a prison setting. The Chicago School of Professional
Psychology. Dissertation Abstracts International. 6 7(1 0-B).

76

3 1293 03063