• unlimited access with print and download
    $ 37 00
  • read full document, no print or download, expires after 72 hours
    $ 4 99
More info
Unlimited access including download and printing, plus availability for reading and annotating in your in your Udini library.
  • Access to this article in your Udini library for 72 hours from purchase.
  • The article will not be available for download or print.
  • Upgrade to the full version of this document at a reduced price.
  • Your trial access payment is credited when purchasing the full version.
Buy
Continue searching

Validity of the juvenile sex offender assessment protocol revised (J-SOAP-II) with a multiethnic sample of juvenile sex offenders released from juvenile justice commission placements

ProQuest Dissertations and Theses, 2011
Dissertation
Author: Ricardo Martinez
Abstract:
The Juvenile Sex Offender Assessment Protocol Revised (J SOAP II) is currently the most widely used and researched risk assessment instrument specifically designed to aid in the evaluation of the risk of sexual recidivism for known juveniles sex offenders (JSO). However, the instrument has been the subject of only a handful of empirical studies and support for its predictive validity has been inconsistent. Furthermore, despite evidence that the predictive validity of risk measures may vary across racial/ethnic groups, the instrument has been primarily tested only with Caucasian JSO. The present study employed a retrospective file review design to complete J SOAP II protocols on 156 JSO who had been released from two Juvenile Justice Commission placements; a residential sex offender treatment center (PRTC) and the state's juvenile correctional center (JTSB). The follow up period was 10 years and state juvenile and adult records of rearrests were used for outcome data. It was hypothesized, that the predictive accuracy of the overall scale would be good for sexual and general reoffense, but that there would be racial/ethnic and study site differences in performance of J SOAP II subscales. The J SOAP II displayed had good internal consistency and interrater reliability. However, contrary to expectations the J SOAP II Total score consistently underperformed (i.e., AUC < .70) with sexual and general reoffense with AUCs of .64 and .60, respectively for the entire sample. The instrument also performed poorly for sexual and general reoffense with participants from PRTC (AUC = .65 and .58, respectively) and wit those from JTSB (AUC = .62 and .60, respectively). Analyses across racial/ethnic subgroups revealed similarly poor performance for sexual and general reoffense with African Americans (AUC = .68 and .62, respectively) and Latino participants (AUC = .21 and .47, respectively), but somewhat higher, albeit nonsignificant, performance with Caucasian participants (AUC = .80 and .64, respectively). Analysis of subscale performance was generally poor with somewhat more promise for the dynamic subscales. Overall, the study did not support the use of the J SOAP II with the high risk JSO studied or with non-Caucasian JSO. Study limitations, clinical implications, and contextualization of the study's results in view of the existing literature are provided.

TABLE OF CONTENTS

CHAPTER I

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

1

Thesis

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

1

Review of the Literature

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

9

Juvenile Sex Offenders (JSO)

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

9

Demographic characteristics of JSO

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

11

Juvenile sex of fender subgroups

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

13

Offense - specific characteristics

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

16

Recidivism.

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

19

JSO recidivism risk factors

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

25

Adult sex offender recidivism risk factors

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

31

Risk Assessment

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

33

Purely clinical risk assessment ................................ ................................ ..............

38

Actuarial risk assessment

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

40

Empirically guided clinical risk assessment

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

45

Risk Assessment with JSO ................................ ................................ ........................

48

The Juvenile Sex Offender Assessment Protocol (J - SOAP)

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

50

Juvenile Sexual Offense Recidivism Risk Assessment Tool – II (J - SORRAT - II).

58

Risk Assessment Matrix (RAM)

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

59

Estimate of Risk of Adolescent Sexual Recidivism (ERASOR)

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

61

Purpose and Rationale ................................ ................................ ...............................

65

Conceptual Hypotheses

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

68

CHAPTER II

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

72

iii

Method

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

72

Participants

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

72

Exclusion criteria

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

75

Measures

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

76

The Juvenile Sex Offender Assessment Protocol Revised

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

76

Procedure

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

82

Outcome data

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

84

Informed consent

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

85

Operational Hypotheses

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

85

CHAPTER III

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

9 0

Results

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

90

Descriptive Statistics

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

90

Racial/ethnic study group descriptive statistics

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

95

J - SOAP - II score descriptive stat istics

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

98

Racial/ethnic group differences in J - SOAP - II scores

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

99

Reliability analyses

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

100

Inferential Statistics

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

102

Hypothesis 1a.

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

102

Hypothesis 1b. ................................ ................................ ................................ .....

103

Hypothesis 1c.

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

103

Hypothesis 2a.

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

104

Hypothesis 2b. ................................ ................................ ................................ .....

110

Hypothesis 3a.

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

116

iv

Hypothesis 3b. ................................ ................................ ................................ .....

116

Hypothesis 3c.

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

117

Hypothesis 4a.

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

118

Hypothesis 4b. ................................ ................................ ................................ .....

118

Hypothesis 4c.

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

119

Hypothesi s 5a.

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

121

Hypothesis 5b. ................................ ................................ ................................ .....

122

CHAPTER IV

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

125

Discussion

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

125

Participants

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

126

Study S ubgroup D ifferences

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

128

Reliability of the J - SOAP - I I

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

129

Subgroup Differences in J - SOAP - II Scores

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

129

Predictive Accuracy of the J - SOAP

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

133

Predictive Accuracy of the J - SOAP - II Across Study Sites

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

135

Predictive Accuracy of the J - SOAP - II Across Racial/Ethnic Groups

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

137

Analyzing the J - SOAP - II Using Survival Analysis

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

138

Study Limitations

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

139

Clinical Implications

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

142

Conclusions

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

143

References

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

146

Appendix A

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

155

Juvenile Sex Offender Assessment Protocol – Revised (J - SOAP - II) Scoring Form ....

155

v

Appendix B

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

157

Let ter of Support NJ State Juvenile Justice Commission

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

157

Appendix C

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

159

Letter of Support from the NJ State Judiciary

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

159

Appendix D

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

161

Fordham University IRB Approval

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

161

ABSTRACT

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

163

VITA

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

165

vi

L IST OF T ABLES

Table 1:

Descriptive Statistics for Entire Sample and by Study Site

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

91

Table 2:

Offense Related Data by Study Site

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

93

Table 3: Descriptive Statistics By Racial/Ethnic Study Group

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

96

Table 4: Offense Related Data By Racial/Ethnic Study Group

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

97

Table 5:

J - SOAP - II

Total And Subscale Descriptive Data

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

99

Table 6:

J - SOAP - II

Comparison Of Means Between Racial/Ethnic Groups

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

99

Table 7 : J - SOAP - II

Scale and Interrater Reliabi lity Data (k = 2, N = 25)

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

101

Table 8 : J - SOAP - II

Comparison of Means between Study Sites

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

103

Tabl e 9 : J - SOAP - II

Predictive Accuracy with Sexual Reoffense for the Total Sample (N = 156)

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

105

Table 10 : J - SOAP - II

Predictive Vali dity with Sexual Reoffense for PRTC and JTSB S ubsamples

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

106

Table 11 : J - SOAP - II

Predictive Accuracy with Sexual Reoffense for Racial/Ethnic S ubsamples

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

108

Table 12 : J - SOAP - II

Predictive Validity with General Reoffense for the Total Sample, N = 156

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

111

Table

13 : J - SOAP - II

Predictive Validity with General Reoffense for the PRTC/JTSB S ubsamples

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

113

Table 14 : J - SOAP - II

Predictive Valid ity with General Reoffense for Racial/Ethnic S ubsamples

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

114

Table 15 : Summary of H ypotheses and R esults

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

124

vii

AKNOWLEDGEMENTS

The list of individuals who m

I credit for helping me to complete this project is too long to note here. Some of you provided guidance, support, and inspiration , while others have given me love, laughter, and needed distraction. I deep ly

appreciate your pl ace in my life

and your role in getting me to this point .

Several noteworthy individuals

were critical to the specific task s

of this research . My advisor, Barry Rosenfeld, whose intellectual fingerprints are on every page that follows and who se

patience,

encouragement, and empathic nudging resuscitated this important project on many occasions . I hope you never have to type another subject line that starts with ―where are you…‖ or ―are ever going to finish this thing…‖ My dissertation committee, whose thou ghtful feedback and high standards have helped produce a project that I will be proud to share with my field. My research assistant and work partner, Jacqueline Flores, for giving up several Saturdays to collect data with me in a shed in the woods of South

Jersey, and smiling for most of the time.

The staff of the N.J. Juvenile Justice Commission and the N.J. State Judiciary for granting permission for this research and for literally, and figuratively, opening institutional doors for the collection

of the data for this project. In particular, Barbara Chayt, Michael Alloise, Mark Davies, Mike Brady, John Soper, Leticia Callender, Jason Fleming, and J ohn V anloan . I am sure there are others. The work you do is important .

M y mentor, Dr. Denise Johns on, who taught me that offender work is abuse p revention work

and for first planting in me the idea that I could and should become a psychologist.

viii

My mother, Angeles Martinez, who is still raising me and who allowed me to focus on this work knowing my children were in great hands . Similarly, my father, Julian Martinez, who taught me to always be proud of what I have to offer

and that I should always offer what I have .

My many friends and family for being there when I needed you and for not being there ,

when you knew I had to write.

My children Jasmine, Jade, and Jaggie for showing maturity beyond your years in letting daddy leave to do school work on many weekends. One tear at the door would have broken me .

Finally, m y wife Karen Beltran, who has n ot only supported , financed, and tolerated this endeavor, but has never lost faith that I could finish. Thank you immensely, for making me the beautiful family that drives me and for every day inspiring me to be a better man.

R esearch like the one that fo llows is relatively rare, because it takes the dedication and work of many special individuals . It was truly my honor to have collaborated with all

of you .

1

CHAPTER I

Thesis

Statistics show that juveniles (i.e., individuals aged 12

to 18) a ccount for 20% to 50% of all arrests for sexual offenses (Federal Bureau of Investigation [FBI] , 20 10 ; Zolondek, Abel, Northey, & Jordan, 200 1 ). S ome of these juvenile sexual offenders ( note : the commonly used acronym JSO

will be used throughout this disse rtation in place of the phrase juvenile sex offenders to avoid repetition ) engage in multiple sexual assaults and often display escalating patterns with respect to offense severity and frequency ( Becker, 1988; Becker & Abel ,

1985;

Becker, Cunningham - Rathne r, & Kaplan, 1986 ; Wieckowski, Hartsoe, Mayer, & Shortz, 1998). In the largest study of JSO to date, 90 clinicians across 30 states provided data on 1616 JSO (Ryan, Myoshi, Metzner, Krugman, & Fryer, 1996). The study found that one

third of the juveniles’ sex offenses involved vaginal or oral penetration, and threats and violence were often present.

A recent meta - a nalysis of 63 studies found a sexual reoffense rate

of 7.08% (range:

0% to 18%, SD = 3.9%)

and a general reoffense rate of 43.4% (range: 10.7% t o 79.9%, SD = 18.9%)

for juvenile sex offenders

(Caldwell, 2010) .

S everal definitional and methodological issues account for the wide range found in

rates of reoffending among JSO. These include low rates of disclosure of sexual abuse, sampling inconsisten cies across studies, follow - up times that vary across studies, and sensitivity differences in the outcome measures used (e.g., re arrest versus re conviction). Several recent reviews of the extant literature have estimated the rate of sexual reoffense for kn own JSO to range from 0% to 37%, with rates of general or violent reoffending by JSO across studies rang ing

from 2% to 80% (Caldwell, 2002 , 2010; Parks ,

200 4 ; Worling &

Långström, 2003).

2

Although exact rates of reoffending are not available, there has been

growing interest in identifying the JSO most likely to reoffend after having been arrested or sanctioned.

Significant strides have been made in the assess ment of

risk of reoffense in adult sex offenders. Currently, there are several well - established mea sures for assessing the risk of sexual and violent recidivism

by adults

that have excellent reliability and good predictive accuracy (Barbaree, Seto, Langton, & Peacock, 2001; Hanson & Harris, 2001; Hanson & Thornton, 2000; Harris

& Rice , 2003). However, o nly within the past decade have researchers develop ed

risk assessment techniques for JSO. In fact, only a handful of juvenile sex offender risk assessment instruments have been the subject of peer - reviewed examination : the Juvenile Sex Offender Assessment Protocol (currently the J - SOAP - II ; Prentky & Righthand, 2003), the Estimate of Risk of Adolescent Sex Offender Recidivism (currently the ERASOR 2.0; Worling & Curwen, 2001),

Juvenile Sexual Offense

Recidivism Risk Assessment Tool - II

(currently the JSORRAT - II; Epperson & Ralston, 2006), and t he Risk Assessment Matrix (RAM;

Richardson, 1995, as cited in Christodoulides, Richardson, Graham, Kennedy, & Kelly, 2005). These four

instruments together have been the focus of only a handful of published studies evalu ating their reliability or predictive validity.

The most well - known of the four

instruments is the J - SOAP - II

(Prentky & Righthand, 2003) . Preliminary studies have found support for the 28 - item checklist’s reliability and factor

structure ( Martinez , Flore s, & Rosenfeld , 2007; Parks & Bard, 2006; Prentky

et al., 2010; Righthand et al., 2005 ; Viljoen et al., 2008). Ri ghthand et al. (2005) found interrater reliability coefficients

( i.e., Intraclass Correlation

Coefficients; ICC)

for the J - SOAP’s four subscale s that ranged from

.80 and .91. Internal consistency

3

(Chronbach’s coefficient alpha ;

α )

for the J - SOAP’s subscales ranged from .64 to .95. Additionally, two unpublished studies

(Hecker, Scoular, Righthand, & Nangle, 2002 ; Waite, Pinkerton, Wieckowski, McGarvey, & Brown, 2002) , cited in the J - SOAP - II

manual, provided

mixed support for the validity of the original J - SOAP and its subscales

(Prentky & Righthand, 2003) . Unfortunately, the low rate of detected reoffense in both studies allowed for only an assessment of the scale’s concurrent validity with an established case management instrumen t and clinician - determined placement settings but not with recidivism.

Martinez

et al.

(2007) also found adequate interrater reliability for the J - SOAP - II

in a sample of mostly ethnic minority JSO treated at an outpatient program, with ICC s

ranging from .70 to .86 for the Total score and subscales. Internal consistency c oefficients were also adequate, with α

r anging from .76 to .87. Receiver Operating Characteristic (ROC) curve analyses demonstrated adequate predictive accuracy rates for the J - SOAP - II

Tot al score, with A rea U nder

the C urve (AUC) estimates of .76 for general reoffense and .78 for sexual reoffense. Additionally, mixed, but overall positive, support was found

for the J - SOAP - II

subscales at predicting sexual

and general

reoffense, as well as t reatment compliance.

On the other hand, a study of 156 male JSO ( m ean age = 18.6) from Oklahoma, followed into adulthood after release from a secure residential sex offender treatment program ,

did not support the predictive accuracy of the J - SOAP - II

(Par ks & Bard, 2006). Participants were mostly

of Caucasian (62.8%) and African - American

(18.6%)

de s cent. The study used clinical and administrative file data to cod e

the J - SOAP - II

and reconviction for a sexual offense as an outcome measure. The study found go od interrater

4

reliability for a subset of 32 cases independently coded by two clinicians, with ICC s

ranging from of .81 to .95, and adequate internal consistency for the Total and subscale scores, with α

coefficients ranging from .77 to .90. Parks and Bard (2006) found support for the J - SOAP - II ’s Scale II (Impulsive/Antisocial Behavior) for sexual reoffense ,

β

= .30, SE

= . 15 , p = . 04, OR

= 1.35, 95% CI [1.01, 1.80]. However, no other J - SOAP - II

scale pre dicted sexual or nonsexual reoffense at rates better than chance. Of note, an inverse relationship was found

between the Scale I ( Sexual Drive /Preoccupation) and nonsexual reoffense, as lower scores were associated with higher rates of reoffense .

In anot her study, the predictive accuracy of the J - SOAP - II

was no better than chance when tested with a multiethnic sample of 128 JSO followed for 1.5 to 13 years after they were release d

from an outpatient treatment program in Texas (McCoy, 2007). Interrater rel iability was variable in this study, with ICCs between the first author and three research assistants of .99, .8 2 , and .39

for the J - SOAP - II . E ven wider variability was observed between the raters across subscales, with ICCs ranging from .00 to .92. Adequa te internal consistency was found in pre -

and post - treatment ratings of the J - SOAP - II

T otal score, with

α

coefficients of .79 and .83 ,

respectively. Pre -

and post - treatment ratings for the J - SOAP - II

subscales were modest ,

with α

coefficients of .71 and .70 (for pre -

and post - treatment ,

respectively) for Scale I ( Sexual Drive /Preoccupation) and .76 and .78 fo r Scale II (Impulsive/Antisocial Behavior). Similarly, Scale III’s (Intervention) pre -

and post - treatment α

coefficients were .76 and .88, while Scale IV ( Community Stability/Adjustment)

had poor internal consistency at the pre -

and at the post - treatment r ating stage ( α

= . 36 and .42 ,

respectively). McCoy (2007) found that only S cales II (Impulsive/Antisocial Behavior) and III (Intervention) were correlated with

5

treatment failure, with modest, but significant point biserial correlation coefficients,

r pb

= . 20, p

< .05, and .21, p

= .05, respectively. Scale II (Impulsive/Antisocial Behavior) was also correlated with nonsexual reoffense, r pb

= . 20, p

< .05. However, further analyses using ROC analyses found no support for the J - SOAP - II

post - treatment Total sco re in predicting sexual

reoffense (AUC = .60) or nonsexual reoffense (AUC = .50).

Similarly,

a more recent study by Viljoen et al. (2008) failed to support the predictive validity of the J - SOAP - II

for sexual reoffense . The researchers coded the J - SOAP - II

u sing the clinical and administrative files of 169 JSO released from a nonsecure residential treatment program in a medium - sized Midwestern U.S. city. Similar to previous studies, the researchers obtained adequate interrater reliability for the J - SOAP - II ’s Total score (ICC = .84). However, the participants’ J - SOAP - II

Total score failed to predict sexual

reoffending

(AUC

= .54 )

and

performed only slightly better than chance (AUC = .63) at predicting serious nonsexual violent reoffending. Furthermore, none of the J - SOAP - II

subscales predicted sexual

reoffending. Only Scale II

(Impulsive/Antisocial Behavior) consistently performed better than chance at predicting ―nonsexual,‖ ―serious nonsexual,‖ and ―any‖ reoffending with significant but generally poor AUC esti mates of .61, .67, .64, respectively. Scale IV (Community Stability/Adjustment) also had a weak

but statistically significant AUC of .63 when predicting ―serious nonsexual violent reoffenses.‖

Viljoen et al. (2008) also divided their sample into those JSO

aged 12

to 15 and those aged 16

to 18 to examine the J - SOAP - II’s predictive accuracy across age groups. The researchers found that the J - SOAP - II underperformed with younger and older study participants, with AUC estimates for sexual and nonsexual reoffens e of .59 and .53

and

6

.54 and .58 respectively. Similarly, the scale performed poorly for serious nonsexual reoffense, with 12 -

to 15 - year - old participants, where an AUC of .44 was found. Only f or serious nonsexual reoffense ,

by 16 -

to 18 - year - old

participa nts ,

did the J - SOAP - II

Total score have significant, although still modest, predictive accuracy, with an AUC of .71.

Significant methodological differences may have led to the pronounced variability i n the J - SOAP - II ’s performance among

the se studies. For example, the Martinez et al.

(2007) study used a sample that was comprised of predominantly inner - city racial/ethnic minority participants who were deemed to be at

sufficiently

low risk as to permit them to be treated in the community. On the other han d, the sample studied by Viljoen et al.

(2008) ,

which was from the Midwest ern

United States, w as

almost exclusively Caucasian and had been deemed to present enough risk of reoffense as to require residential treatment. Furthermore, while the Viljoen et al .

(2008) study used rearrest as the outcome measure, reoffense data in the Martinez et al. (2007) study came from various sources ,

including parents, self - report, child protective services, school officials, and parole/probation officers, thus including som e reoffenses for which the participants may have never been charged

with a crime .

In addition to mixed evidence of the J - SOAP’s reliability and ability in

predict ing sexual

and general recidivism, small sample sizes

(with the exception of Prentky

et al. ,

20 10 ) and low base rates of reoffense among the samples studied thus far have not allowed for thorough examination of the scale’s predictive utility.

In order to establish the robustness of a risk assessment instrument, replication across different samples

and subgroups of offenders is necessary (Hanson & Thornton, 1999). Factors such as age and racial/ethnic background may have unknown interaction effects with known reoffense

7

risk factors ,

thereby affecting the predictive validity of risk assessment measur es. For example, when Viljoen et al. (2008) limited the ir

sample of 196 JSO to only those aged 16 to 18, the J - SOAP - II

Total score generated a higher AUC

that

was significan tly better than chance

at predicting several outcome measures. Specifically, the J - SOAP - II

Total score was a significant predict or of

―sexually aggressive behavior‖ (AUC = .67), ―nonsexual/aggressive behavior‖ during treatment (AUC = .69), and post - treatment ―serious nonsexual violence‖ (AUC = .71). To date, no studies comparing the J - SO AP - II ’s predictive accuracy across racial/ethnic subgroups of JSO have been conducted . The absence of evidence of comparable validity across ethnic groups is

a significant concern ,

considering findings

from the adult sex offender literature

that indicate t hat the predictive accuracy of risk assessment instruments may vary across racial/ethnic groups (Långström, 2004; Schwalbe, Fraser, Day, & Cooley, 2006).

The present study used a retrospective review of the mental health records and criminal history data

of JSO released from two NJJJC facilities to assess the J - SOAP - II ’s reliability and ability to predict sexual and nonsexual (i.e., general) reoffense. The primary investigator coded the J - SOAP - II , demographic information, and reoffense data f o r

167 JSO re leased from a NJJJC

correctional facility and a residential treatment program in New Jersey using clinical records and state judiciary juvenile and adult arrest databases .

I n addition to assessing the scale’s overall

ability to predict reoffense, the pre sent study assessed whether the J - SOAP - II

is equally adept at predicting sexual and general reoffense with African - American , Caucasian, and Latino JSO. Finally, the current study used a sample of JSO from a correctional placement and one sample from a resi dential

8

treatment facility, allowing for a comparative assessment of the instrument’s predictive accuracy with two presumably qualitatively different types of JSO groups.

It was hypothesized that the J - SOAP - II ’s Total score would display adequate predict ive accuracy for sexual and general reoffense

with the entire sample (i.e., all subgroups combined). However, based on previous research (Martinez et al., 2007 , Prentky et al., 2010; Rajlic

& Gretton,

2010 ) ,

it wa s expected that Scale I (Sexual Drive/Preoc cupation) w ould

perform significantly better with Caucasian offenders than with African - American

and

Latino youth and better with JSO released from residential treatment

than with participants who were released from the correctional facility. On the other hand, Scale II (Impulsive/Antisocial Behavior) w as expected to perform better at predicting sexual and general reoffense with African - American

and Latino participants compared to Caucasians and with those released from the correctional facility

compared to

those released from the residential treatment center.

The following review of the literature will provide information regarding the prevalence of juvenile sexual offending

and

the demographic and offense - specific characteristics of JSO and will review wh at is known about risk factors for reoffense and rates of recidivism for JSO. Next, the development and status of various techniques and procedures for assessing the risk of recidivism among JSO is detailed, with special emphasis on the arguments for and a gainst the use of clinical judgment, actuarial instruments, and empirically guided risk assessment tools like the J - SOAP - II . The chapter concludes with a detailed review of several studies examining the reliability and predictive validity of the J - SOAP - II ,

while illuminating significant gaps in the extant literature on the predictive utility of the J - SOAP - II .

9

Review of the Literature

Juvenile S ex O ffenders (JSO)

Until the late 1980s juvenile sexual offen ses

w ere

considered rare and largely an issue of male adolescent experimentation (Becker, 1988). The sexually abusive behavior of juveniles was usually ignored, minimized, or denied (Ryan et al., 1996). However, the past two decades have s een marked change in the appraisal s

of th is

problem by the public, the legal system, and the scientific community. It is now understood that the sexually abusive behavior perpetrated by juveniles is a complex problem that poses a significant risk to the juvenile’s developmental trajectory and to community safety (Wieckowski e t al., 1998).

According to the United States Bureau of Justice Statistics (Greenfeld, 1997), juveniles (i.e., individuals aged 12 - 18) constituted roughly 16% of all the arrests for rape and 17% of other sexual offenses in 1995. In 200 9 , juveniles accounted

for 1 2 , 9 52 (1 7 %) of the 76 , 864

arrests made for sexual offenses, excluding prostitution (FBI, 20 10 ). Additionally, although juveniles are responsible for only 4% of violent sexual assaults against adult women, they account for 35.6 % of all arrests for off en ces against minors (Snyder, 2000).

In the largest study of JSO conducted to date, Ryan et al. (1996) surveyed 90 clinicians

across 30 states ,

finding that 35% of offenses committed by the 1616 JSO studied involved vaginal or oral penetration. Verbal coe rcion, threats, and physical force were present in 57%, 25%, and 31% of cases ,

respectively. Several studies using self - report questionnaires and clinician assessment have found that many JSO are never formerly charged for sexually abusive behaviors ( Spacc arelli, Bowden, Coatesworth, &

10

Kim, 1997; W orling & Curwen, 2000; Zolondek et al.,

2001) . Relying on clinician and offender self - report data introduce s

the possibility of bias due to s ocial desira bility on the part of offenders as well as bias due to clini cian’s positive regard for their

own clients. However, these potential contaminants to the objectivity or accuracy of the data collected would likely result in the underestimation of the level of psychopathology and deviant behavior found among these parti cipants. As such, data obtained via offender self - report or from their clinicians are likely to be underestimations of the true extent and severity of their behavior.

Viewing the problem retrospectively, about 50% of incarcerated adult

sex offenders adm it to having committed sexual offenses before the age of 18 (Abel, Mittelman, & Becker, 1985; Abel & Rouleau, 1990), suggesting that some JSO will continue offending into adulthood. There is also evidence that an early onset of sexual offending is associat ed with higher rates of adult sexual reoffending (Hanson & Bussiere, 1998 ;

Prentky

et al., 2010 ).

Research on the behavior of JSO that is based on legal charges, clinician ratings, and sex offender self - report suggest s

that many JSO have engaged in multip le sexual assaults and that some display escalating patterns with respect to severity and frequency (Becker & Abel ,

1985; Wieckowski et al., 1998).

S tudies documenting the offense - related behaviors of juveniles show that the offenses of juveniles are

often

characterized by high levels of force , penetration, manipulation, and deviance ( Miranda & Corcoran, 2000; Ryan et al., 1996; Wieckowski et al., 1998; Zolondek et al., 2001). However, JSO - specific research continues to lag behind the literature on adult se x offenders .

11

Demographic characteristics of JSO .

To

date only two large - scale cross - sectional studies have reported demographic data on JSO

samples . Ryan et al. (1996) conducted a descriptive study, using background and clinical information gathered from

90 treatment pro viders from across 30 states. Clinicians were asked to complete four different structured questionnaires that elicited demographic information and clinical impressions on JSO being evaluated. The final sample of 1616 JSO was 97% male

and h ad

a modal age of 14

( with 90% between 10 and 18 years of age ) . The study sample was a cross - section of the country, with 27.4%

of the sample originating from rural communities , 21% from communities with populations between 15 and

50 thousand, 18.6% from c ommunities with populations between 50 and 250 thousand, 16.1 %

from suburb s of large cities, and 16.9 %

from large cities with populations greater than 250 thousand . Although the racial/ethnic make - up of the sample was not provided,

Full document contains 175 pages
Abstract: The Juvenile Sex Offender Assessment Protocol Revised (J SOAP II) is currently the most widely used and researched risk assessment instrument specifically designed to aid in the evaluation of the risk of sexual recidivism for known juveniles sex offenders (JSO). However, the instrument has been the subject of only a handful of empirical studies and support for its predictive validity has been inconsistent. Furthermore, despite evidence that the predictive validity of risk measures may vary across racial/ethnic groups, the instrument has been primarily tested only with Caucasian JSO. The present study employed a retrospective file review design to complete J SOAP II protocols on 156 JSO who had been released from two Juvenile Justice Commission placements; a residential sex offender treatment center (PRTC) and the state's juvenile correctional center (JTSB). The follow up period was 10 years and state juvenile and adult records of rearrests were used for outcome data. It was hypothesized, that the predictive accuracy of the overall scale would be good for sexual and general reoffense, but that there would be racial/ethnic and study site differences in performance of J SOAP II subscales. The J SOAP II displayed had good internal consistency and interrater reliability. However, contrary to expectations the J SOAP II Total score consistently underperformed (i.e., AUC < .70) with sexual and general reoffense with AUCs of .64 and .60, respectively for the entire sample. The instrument also performed poorly for sexual and general reoffense with participants from PRTC (AUC = .65 and .58, respectively) and wit those from JTSB (AUC = .62 and .60, respectively). Analyses across racial/ethnic subgroups revealed similarly poor performance for sexual and general reoffense with African Americans (AUC = .68 and .62, respectively) and Latino participants (AUC = .21 and .47, respectively), but somewhat higher, albeit nonsignificant, performance with Caucasian participants (AUC = .80 and .64, respectively). Analysis of subscale performance was generally poor with somewhat more promise for the dynamic subscales. Overall, the study did not support the use of the J SOAP II with the high risk JSO studied or with non-Caucasian JSO. Study limitations, clinical implications, and contextualization of the study's results in view of the existing literature are provided.