Add To Cart



Selected Readings

Test | Table of Contents

Additional Readings

The Effect of Adolescent Sex Offender Abuse History on Counselor Attitudes. By: Carone, Stacia S.; LaFleur, N. Kenneth. Journal of Addictions & Offender Counseling, Apr2000, Vol. 20 Issue 2, p56, 8p, 3 charts; (AN 2980357)

Counseling students' judgments of adolescent sex offenders with sexual or physical abuse histories were examined. Sexually abused offenders were more desirable as prospective clients than were nonabused offenders. Sexually abused counselors desired to see physically abused offenders as clients over sexually abused offenders. Implications for counseling are presented.
Until recently, therapeutic treatment of adolescent sexual offenders has been lacking. Research has not adequately explored the therapeutic relationship between adolescent sex offender clients and their counselors. Specifically, research has scarcely addressed counselors' judgments about working with adolescent sex offenders. Given the nature of such clients' problem (i.e., sexual acting out), many counselors may be hesitant to work with offenders because such behavior has been judged primarily as a crime and not as a therapeutic issue (Steen & Monnette, 1989). Consequently, counselors' judgments about working with adolescent sex offenders will affect their ability to develop therapeutic relationships with this population (Steen & Monnette, 1989). With increasing numbers of adolescent sexual offenders being reported, counselors will need to explore their judgments and make informed decisions about working with these clients (Farrenkopf, 1992). Counselor judgment has been defined by Thomas, Sayers, Borgers, and Barke (1987) as the counselor's desire to work with a client and the counselor's perception of the client's need for counseling. There are many factors that influence counselors' desire to work with sex offenders and counselors' perceptions of offenders' need for counseling. Such factors may include a history of physical or sexual abuse reported by the client (Bruinsma, 1988). Researchers have found that many sex offenders report sexual or physical abuse histories that may affect the offenders' current offending behaviors. Learning theories (Freeman-Longo, 1986; Steen & Monnette, 1989) support the idea that a sexual or physical abuse history influences offending behaviors. The theories may help counselors develop a rationale for working with sex offenders. The use of these theories may increase the counselor's understanding of the offender as victim. Learning theories also provide counselors with a model from which to work in addressing the cycle of abuse and offending by adolescent clients. Counselors may judge sex offenders with no reported abuse history differently than they will offenders with reported abuse histories because, in the first case, a rationale for offending behaviors may not be determined. Offenders lacking a reported abuse history may be perceived as criminals without victim status. Some theories explaining offending behavior by nonabused offenders include congenital or acquired brain damage (Hendricks et al., 1988) and in utero exposure to high levels of opposite sex hormones (McConaghy, 1993). Such problems are chronic, biologically based, and difficult to treat. Other explanations for offending behaviors are extracted from criminal justice theories. Counselors may see the need for more intrusive intervention for nonabused offenders, such as medication or incarceration, or both.
The physical or sexual abuse history of counselors may also affect their judgments of adolescent sex offenders because counselors have been reported to feel traumatized by working with sexual offender clients (Allen & Brekke, 1996). As a result of counselor experiences of sexual abuse, Mitchell and Melikian (1995) have revealed that counselor countertransference reactions to sexual offenders may create personal difficulties in working with this population.
Research in counselors' judgments of adolescent sex offender clients is scarce. Accordingly, this study was designed to identify differences in counselor judgments of adolescent sex offender clients with reported abuse histories and of adolescent sex offender clients without reported abuse histories. The following hypotheses were tested:

  1. Student counselors' desire to work with an adolescent sex offender with a physical or sexual abuse history will be greater than student counselors' desire to work with an adolescent sex offender with no reported abuse history.
  2. Student counselors' opinions of the need of a physically or sexually abused adolescent sex offender for counseling will be greater than counselors' opinions of a nonabused adolescent sex offenders' need for counseling; more specifically, it was hypothesized that student counselors' desire to work with the client and student counselors' opinions of the client's need for counseling would be greater for the sexually abused adolescent sex offender than for the physically abused adolescent sex offender.
  3. Student counselors with a self-reported history of sexual abuse will have different levels of desire to work with each of the three client types.

METHOD

Design

The study used an experimental analogue design. Three case histories of fictional sex offender clients represented the treatments. Treatment assignment was random in an attempt to control for influential institutional program and participant differences.

Participants

Two-hundred thirty-six counselors-in-training who were enrolled in one of eight counselor education master's degree programs located in two mid-Atlantic states were the participants in the study. The student counselors were asked to respond to demographic questions in the form of checklists regarding their gender, race, history of physical abuse, history of sexual abuse, and level of counseling experience. A cover letter indicated that respondents could opt to decline answering any or all of the questions. The counseling students were in the second year of their master's degree programs and varied in their reports of counseling experience from none to more than 1 year working in the field. The participant group consisted of 54 male- and 182 female counseling students. Of the sample, 206 Caucasian-, 20 African American-, 1 Hispanic-, and 6 Asian counselors participated in the study; 3 participants reported their race as "other."

Instruments

The Counselor Response Form (CRF; Borgers, Thomas, & Van Loon, 1980) was used to determine the counselors' judgments of adolescent sex offender clients with various histories including sexual abuse, physical abuse or no abuse. The CRF reports counselor judgment that consists of two dependent variables: the degree to which the counselor believes that the client needs help (need score) and the degree to which the counselor desires to work with the client (desire score). The form consists of 12 statements presented randomly. Five statements measure the counselor's perception of the client's need for counseling. The internal consistency for these items was .94 in a pilot study and .92 in previous research (Thomas et al., 1987). The counselor's desire to see the client is measured by five other statements. The internal consistency for these responses was .92 in both the pilot study and in previous research (Thomas et al., 1987). The unscored items are used as fillers. The statements are rated on a 5-point Likert-type scale from strongly agree (1) to strongly disagree (5), and responses are based on counselor self-ratings regarding a presented case study.

Treatments

Three case histories were constructed presenting different client personal abuse histories. One of three written case histories was presented to each participant. Each case history was approximately 75 words long. The client presented in each case history was a fictional, 15-year-old boy because many offenders are male. The offender presented in each case history performed vaginal intercourse with a 10-year-old girl on two occasions. Two to three lines embedded in the case history discriminated between the offender case histories on the dimension of offender abuse history. Each case included the offender's report of sexual abuse, physical abuse, or no abuse.
Various professional counselors reviewed the case histories, and evidence of content validity was obtained. Criterion validity was addressed by the researchers through comparisons of the fictional case histories with real offender case histories. Researchers conducted t tests on the CRF responses of professional counselors who were asked to respond to the case histories based on preassigned judgments of the different types of offenders. The professional counselors were asked to complete the CRF in response to the case histories--as if they desired to see the client or as if they did not desire to see the client--for each of the three client types. They were similarly assigned judgments for their perception of each of the client's need for counseling. The process permitted a determination of the construct validity of each case history and revealed significant differences at the .01 level between high and low mean ratings for all of the case histories. The counselors' responses to the CRF corresponded to the attitudes they were preassigned toward each case history; therefore, the instrument had evidence of construct validity. The results also revealed that the CRF discriminates between high and low scores of both need and desire for each of the three case histories.

Procedure

Instructors of master's level counseling courses distributed the data collection packets to students in one of their courses. Equal numbers of each case history were randomly distributed in the packets received by each instructor. Each packet contained an informed consent form, a directions page, a demographic questionnaire, the CRF, and one of the three case histories.
Counseling students who agreed to participate in the study completed the informed consent form and the demographic questionnaire, read the case study, and responded to the CRF items. The participants then placed the completed materials in the packet envelope and sealed the envelope. The sealed envelopes with the completed forms were returned to the researchers.

RESULTS


A one-way of analysis of variance (ANOVA) was conducted on the CRF desire and need scores, to determine if there was a significant difference among the student counselor ratings of abuse histories on a measure of the student counselor's perception of the client's need for counseling and on a measure of the student counselor's desire to work with the client. Mean responses for desire and need scores for each group are represented in Table 1.
ANOVA results, F(2,235) = 3.32, p < .05, revealed that there was a significant difference among the three groups' responses of desire to see the client. To control for the probability of a Type I error, and to make pairwise comparisons of the group means on the dimension of desire score, multiple comparisons using Scheffe's test were conducted with a probability level of .05. Comparisons using a Scheffe's test showed that desire scores of the respondents who read the case history of a sexually abused offender (M = 18.76, SD = 4.14) were significantly higher than for respondents who read the case history of a nonabuse offender (M = 17.04, SD = 4.56), t(233) = 2.58, p = .011.
An ANOVA was conducted to reveal any differences between the desire and need scores of those respondents who reported a history of sexual abuse on the demographic questionnaire and those who did not. Need scores did not differ between those respondents who reported a history of sexual abuse and those who did not report a history of sexual abuse. There was an interaction for desire scores between treatment case history and the report of a personal history of sexual abuse by the participants. The group mean data and analysis summary are displayed in Table 2 and Table 3.
Post hoc comparisons using Newman-Keul's test revealed specific significant differences from the interaction. The student counselors who were sexually abused desired to see the physically abused offender (M = 20.38, SD = 3.54) more than they desired to see the nonabused offender (M = 14.50, SD = 5.72), F(2,229) = 5.88, p < .05. The student counselors who were not sexually abused desired to see the nonabused offenders (M = 17.40, SD = 4.30) more than the student counselors who were sexually abused (M = 14.50, SD = 5.72), F(1,229) = 4.34, p < .05.
Testing of the remaining hypotheses revealed no significant differences. No significant differences were found between the CRF need scores of student counselors responding to clients with histories of abuse versus nonabuse or histories of sexual abuse versus physical abuse.

DISCUSSION


This study provides counselors with information regarding variables that affect counselor judgment of adolescent sexual offenders. Counseling students who participated in this study revealed a greater desire to work with a sexually abused adolescent sex offender than with an adolescent sex offender who was not abused, according to participants' CRF scores. These results could be attributed to the idea that sexually abused offenders have learned offending behaviors. The counseling students may have perceived the sexually abused offender as a victim of abuse that caused him to learn the sexually offending behaviors. Nonabused offenders may be less desirable to work with because a potential rationale for their behavior cannot be determined. In addition, nonabused offenders may be perceived by counselors as victims of medical or biological problems who may be better served by professionals who could provide more medically based types of treatment.
Counselor judgments affect the counseling relationship, and the counselor's desire to work with certain clients is a judgment that cannot be overlooked (Borgers et al., 1980). As the need for juvenile offender counselors grows with the increasing rate of reported sexual offenses (Steen & Monnette, 1989), counselors will need to explore their judgments about working with offender clients in order to make informed decisions about their clients.
Because counseling students desired to see the sexually or physically abused adolescent sex offenders more than they desired to see the nonabused offenders, the development of counseling relationships between counselors and nonabused offenders may be affected (Graham, 1980). Graham found that reports of counselor desire to see a client affect the counselor's prognosis of client treatment and the counselor's predicted length of time the client will need in counseling. Consequently, counselors should consider how the characteristics of nonabused adolescent sex offenders affect the counseling relationship. If counselors do not desire to see a nonabused sex offender, they may need to either refer the client to another counselor or consult with other professionals regarding the treatment of nonabused adolescent sex offenders. In addition, Kearns (1995) found that counselors' feelings toward sexual offenders can lead to professional burnout, and, therefore, counselors should use self-reflection to manage their feelings and respond appropriately.
Analysis regarding the judgments of sexually abused student counselors about the adolescent sex offenders with either sexual, physical, or no abuse histories was computed. Mitchell and Melikian (1995) revealed that the background of counselors might affect their judgments about working with offender clients. In the present study, sexually abused counseling students desired to see a physically abused offender more than they desired to see a nonabused offender. The similarities in the backgrounds of the sexually abused student counselors and the physically abused offenders might have contributed to the results. Because some sexually abused counselors have been reported to experience lack of trust, fear, and guilt toward other people (Emerson, 1988), their objectivity may be questioned in working with sexual offenders (Allen & Brekke, 1996). Counselors should understand how their own backgrounds influence their desire to see adolescent sex offenders and its possible effect on counseling outcome. Such counselor understanding could lead to the counselor's decision to continue working with the client, to consult with another professional, or to refer the client to another counselor. Counselor educators should help student counselors work through their own sexual abuse histories and assist them in reflecting on the effect of the histories on their judgments of sexual offenders.
Counseling students who were not sexually abused desired to see the nonabused offender more than the counseling students who were sexually abused desired to see the nonabused offender. Despite these results, nonabused sexual offender clients may still present the nonabused counselor with difficult issues, such as seduction and intimidation (Allen & Brekke, 1996). Therefore, nonabused counselors should also be aware of ways in which they can maintain objectivity in working with sexual offenders.
The respondents did not differ in their ratings of clients' need for counseling that could be attributed to the client's being an adolescent. Most adolescent sex offenders have less experience performing sex offending activities than have adult sex offenders, and they have shown a lower recidivism rate of sexual offending behaviors as a result of treatment than have adult sex offenders (Steen & Monnette, 1989). Future research should address differences in counselor attitudes toward adolescent sex offenders versus adult sex offenders.
Finally, counselor educators should be instrumental in encouraging student counselors to explore their judgments of adolescent sex offenders and potential effects of those judgments on the counseling relationship. Training programs can help counseling students understand the ways in which their judgments of adolescent sexual offending may affect client outcomes. In addition, student counselors could be given strategies for dealing with clients whom they have little desire to work with and clients whom they believe have little need for counseling.
This study has several limitations. First, analogue methodology strengthens internal validity and experimental precision but limits external validity. Second, the respondents were graduate students, which also limits the generalization of results. Similar research using respondents of varying levels of experience will need to be conducted to address the potential influence of counselor experience levels on counselor attitudes toward sex offenders. Third, because additional analyses using the counselor demographic variable of abuse history were conducted, the sample size of sexually abused counselors is limited. Although the researchers found differences in the counseling students' CRF desire scores, using a larger sample for this group with a history of sexual abuse may detect additional differences.
Much more research will need to be conducted if professionals are to gain a more comprehensive understanding of counselor judgments as they relate to sex offender clients. This exploratory study provides a basis from which to begin more research in this area.

------

 

 

 

 

Emotional variables and deviant sexual fantasies in adolescent sex offenders. By: DiGiorgio-Miller, Janet. Journal of Psychiatry & Law, Summer2007, Vol. 35 Issue 2, p109-124, 16p; (AN 27789074)

 

"The power of the mind, as exhibited in sexual fantasy, plays
a key role in both normal and deviant sexual behavior. . . .
Sexual fantasies serve to enhance sexual arousal." (Wright &
Schneider, 1997). The primary focus in sex offender literature
regarding sexual fantasy and subsequent arousal focuses on
deviant sexual interests and resulting behavior. Looman
(1995) noted that despite the acknowledgment of the
importance of sexual fantasies in working with sex offenders,
"little controlled research has been done in this area." Most
literature that examines deviant sexual arousal focuses on
adult sex offenders (e.g., Abel & Blanchard, 1974). Barbarbee
& Marshall (1988) studied variables such as offense history,
IQ, socio-economic factors, and factors related to sexual
deviance as measured by a penile plethysmography for 35
untreated adult child molesters. Sexual deviance was the only
significant predictor of recidivism.
In addition, a reduction of deviant sexual arousal does show
some predictive ability in post-therapy success than any other
measure in terms of recidivism rates. (Quinsey & Marshall,
1983). Deviant sexual fantasies not only influence future
behavior, but they serve to shape ideas, organize perceptions,
increase cognitive distortions, and increase motivation to
engage in future deviant behavior (Wright & Schneider,
1997). Cortini & Marshall (2001) stated, "Deviant fantasizing
is understood to set the stage for a series of decisions that
progressively increase the individual's risk to re-offend."
Most literature that examines deviant sexual arousal focuses
on adult sex offenders. However there is some research that
examines adults who began having deviant fantasies as youth.
I l l
Abel and Rouleau (1990) found that the majority of the 561
adult sex offenders they studied reported that they had
acquired their deviant sexual interest before the age of 16.
Marshall, Barbaree & Eccles (1991) found in their sample of
129 outpatient child molesters that 29% reported having
deviant sexual fantasies prior to age 20. Although this
literature assists the researcher and clinician regarding
conceptualizing the relationship between deviant sexual
fantasy and resulting arousal and sexual offending, there is a
considerable gap in the research regarding deviant sexual
arousal with adolescent sex offenders.
From a review of the literature, several important emotional
variables emerge as relating to deviant sexual fantasies.
Proulux, McKibben & Lusignan (1996) also found loneliness,
anger and humiliation to be associated with deviant sexual
fantasies in their sample of 30 adult males. Proulux, et al.
(1996) stated, "In rapists and heterosexual pedophiles
negative mood and conflicts coincided with overwhelming
deviant sexual fantasies and increased masturbatory activities
during such fantasies." Swaffer, Hollin, Beckett & Fisher
(2000) found in their sample of 30 adult men that depression
and anger were the most frequently reported negative mood
states prior to offending. Looman (1999) found in his review
of the literature that for adult men, negative mood states
produced deviant sexual arousal. He hypothesized that
deviant emotional fantasies were a "way to cope" with
negative emotional states for sex offenders. He further noted
loneliness and inadequacy were found to be the primary
emotions related to deviant sexual fantasies, primarily for
child molesters (66%). He stated, "An anxious-ambivalent
attachment style may characterize a man who sexually
offends against children." Keenan & Ward (2000) also
theorized that intimacy deficits correspond to the
underpinnings of sexual deviance and offending. In Looman's
1995 article, he found that adult child molesters were more
likely to fantasize about children when in a negative
112 EMOTIONAL VARIABLES
emotional state than when in a positive mood. He stated that
the deviant fantasizes serve an emotional need for the sex
offender. The negative mood states were defined as
depression and rejection. Cortini & Marshall (2001) studied
89 incarcerated sex offenders. They found that negative
emotional states are present before the aggression occurs and
therefore is an important link in relapse prevention. Cortini &
Marshall (2001) found that "Sexual offenders showed
evidence of sexual preoccupation during adolescence and this
preoccupation was related to the latter use of sex as a coping
strategy." They, like other authors, noted that intimacy
deficits and loneliness were related to greater use of sexual
activity as a way to cope with negative emotional states.
The most widely documented emotional motivators for sexual
offending are intimacy deficits and loneliness. Marshall
(1989), Garlick, Marshall & Thorton (1996) and Marshall
(1993) noted loneliness in adult child molesters due to their
failure to maintain intimate relationships. Ward, McCormack
& Hudson (1997) also noted loneliness in adult offenders and
stated that sex offenders "may indirectly seek intimacy
through sex, even if they have to force a partner to
participate." Marshall, Champagne, Sturgeon & Bryce (1997)
found that when therapeutic intervention included ways to
reduce loneliness, that sex offenders achieved enhancements
in both intimacy and self-esteem.
Diamant & Windhoz (1981) noted that loneliness in college
students is related to other personality traits, such as
aggression. Jones (1982), Ward, Hudson, Marshall & Siegert
(1995), Seidman & Marshall (1994) and Check, Perlman &
Malamuth (1985) also noted the relationship between
loneliness and hostility. Smallbone & Dadds (2000) found
that insecure attachment in adult males was associated with
antisocial behavior and aggression. Williams (1983) in his
essay on adolescent loneliness stated that those adolescents
with a medium to high need for control are lonelier. This
need for control may lead to aggressive tendencies.
113
Hostility and anger were also noted as related to sexual
offending. Yates, Barbaree & Marshall (1984) found in their
sample of 24 adults, that anger enhances arousal to forced sex
episodes for rapists when compared to control subjects.
Avery-Clark & Laws (1984) also noted that adult violent sex
offenders responded significantly more to aggressive cues
than nonviolent offenders. Barbaree, Marshall & Lanthier
(1979) also found that aggressive cues did not inhibit arousal
in rapists. Looman (1999) found in his review of the literature
that for adult men, negative mood states produced deviant
sexual arousal. He hypothesized that deviant emotional
fantasies were a "way to cope" with negative emotional states
for sex offenders. He further noted loneliness and inadequacy
were found to be the primary emotions related to deviant
sexual fantasies, primarily for child molesters (66%). He
stated, "an anxious-ambivalent attachment style may
characterize a man who sexually offends against children."
Keenan & Ward (2000) also theorized that intimacy deficits
correspond to the underpinnings of sexual deviance and
offending. In Looman's 1995 article, he found that adult child
molesters were more likely to fantasize about children when
in a negative emotional state than when in a positive mood.
He stated that the deviant fantasizes serve an emotional need
for the sex offender. The negative mood states were defined
as depression and rejection. Cortini & Marshall (2001)
studied 89 incarcerated sex offenders. They found that
negative emotional states are present before the aggression
occurs and therefore is an important link in relapse
prevention. Cortini & Marshall (2001) stated, "Deviant
fantasizing is understood to set the stage for a series of
decisions that progressively increase the individual's risk to
re-offend." They also found that "sexual offenders showed
evidence of sexual preoccupation during adolescence" and
this preoccupation was related to the latter use of sex as a
coping strategy." They, like other authors, noted that intimacy
deficits and loneliness were related to greater use of sexual
activity as a way to cope with negative emotional states.
114 EMOTIONAL VARIABLES
Although this literature assists the researcher and clinician
regarding conceptualizing the relationship between deviant sexual
fantasies and resulting arousal and sexual offending, it has been
conducted entirely upon adults. There is a relative lack of research
regarding deviant sexual arousal with adolescent sex offenders.
Dutton & Newlon (1988) surveyed a sample of eight adolescent
sex offenders and 70% of them reported having sexually
aggressive masturbatory fantasies prior to committing their sex
crimes. The subjects noted feelings of inadequacy and low selfesteem.
Wiechowski, Hartsoe, Mayer & Shortz (1998) suggest
that deviant sexual behavior may begin in early childhood prior to
the onset of adolescence. Daleiden, Kaufman, Hilliker & O'Neil
(1998) examined the sexual histories and fantasies of adolescents
by using the Sexual Fantasy Questionnaire. They found from this
self-report measure that youthful sexual offenders reported fewer
consenting experiences than both nonsexual offenders and
nonoffenders. This study attempts to examine mood states such as
loneliness and hostility as it relates to deviant sexual fantasy in
adolescent sex offenders in both inpatient and outpatient services.
The present study is an attempt to address this lack of research
regarding the role of deviant sexual fantasies in adolescents.
Method
Participants There were 66 adolescent male sex offenders, 33 inpatient,
and 33 outpatients, participants in this study. The inpatient
sex offenders came from residential facilities in
Massachusetts (Stetson School) and New Jersey (Pinelands).
The outpatient sexual offenders came from both urban and
suburban settings in New Jersey.
Regarding inpatient offenders, the mean age was 15.26,
standard deviation was 1.45. The mean number of months in
treatment was 13.48, standard deviations was 10.40. The
mean number of offenses was 16.39, standard deviation was
12.60. Both the inpatient juvenile offenders and the
115
outpatient offenders may have had many offenses with one or
more victims. Specifically, regarding the type of offense:
there were 25 participants that reported fondling, 25 oral
penetration, 18 anal penetration, 11 vaginal penetration, 6
frottage, 12 for exposing, and 1 for "other". The mean
number of victims was 1.39, standard deviation was .496.
Regarding outpatient offenders, the mean (M) age was 15.92,
standard deviation (SD) was 1.83. The mean number of
months in treatment was 12.73, standard deviations was 9.21.
The mean number of offenses was 4.33, standard deviation
was 3.24. Specifically, regarding the type of offense: there
were 27 participants that reported fondling, 10 oral
penetration, 13 anal penetration, 11 vaginal penetration, 0
frottage, 3 for exposing, and 2 for "other". The mean number
of victims was 1.26, standard deviation was .448. Table 1
describes the descriptive statistics regarding the inpatient and
the outpatient demographic data.

 

 

-------

 

 

 

Risk assessment with adolescent sex offenders. By: Christodoulides, T. E.; Richardson, G.; Kennedy, P. J.; Graham, F.; Kelly, T. P.. Journal of Sexual Aggression, Jan2005, Vol. 11 Issue 1, p37-48, 12p, 3 charts; DOI: 10.1080/13552600410001697848; (AN 15610084)

Introduction
Sexual abuse perpetrated by children and young people has received increasing attention over
the past decade (Masson & Erooga, 1999). This is due to ‘‘an increased sensitivity to the
manifest harm caused by these youngsters to their victims, and an increased awareness of the
incidence of juvenile sexual aggression’’ (Prentky, Harris, Frizell, & Righthand, 2000).
The incidence of juvenile sexual aggression is indeed considerable: 23% of the 6400
individuals cautioned or found guilty of sexual offences in the UK in 1997 were between 10
and 21 years of age (Home Office, 1998). This is consistent with other estimates. Becker,
Kaplan, Cunningham-Rathner, and Kavoussi (1986), for example, estimated that 20% of all
rapes and 30_/50% of child molestations are committed by adolescent males. Reviews
completed by researchers such as Openshaw et al. (1993) suggest that children and
adolescents account for approximately 25_/33% of sexual abuse perpetrators. The literature
also indicates that a large percentage of adult sexual offenders committed sexual offences in
adolescence. Abel, Mittelman, Becker, Rathner, and Rouleau (1988) report that over 50% of
*Corresponding author: Thomas Kelly, Nothern Forensic Mental Health Service for Young People, St Nicholas
house, St Nicholas Hospital, Gosforth, Newcastle Upon Tyne, United Kingdom NE3 3XT. Tel: _/44 191 223 2226.
Fax: _/44 191 223 2228. E-mail: thomas.kelly@dsl.pipex.com
ISSN 1355-2600 print/1742-6545 online # 2005 National Organisation for the Treatment of Abusers
DOI: 10.1080/13552600410001697848
all adult sex offenders disclose some form of sexually deviant behaviour and/or interest prior
to the age of 18. Similar incidence figures have been reported by Marshall, Laws, and
Barbaree (1990).
Given that even the lowest estimates of incidence are considerable, there are clear
benefits to identifying those juvenile offenders who represent a high risk of sexual reoffending.
Identifying those at risk of developing a pervasive pattern of persistent sexual offending into
adulthood would serve to inform a number of key decisions: where to appropriately place the
perpetrator; whether to seek a referral to a multi-agency public protection management
meeting; how to minimize future risks, whilst simultaneously helping to promote the
development and welfare of the abuser; deciding the most appropriate form of intervention;
and identifying educational provision for the perpetrator and allocating resources (Epps,
1999).
It is not surprising that the adolescent sexual offender risk assessment literature suffers
from the same methodological weaknesses as the adult literature. These weaknesses include
reliance upon reconviction rates (an accepted significant underestimation of true recidivism,
Home Office, 1988), lack of untreated matched samples, small sample sizes, and limited
follow up periods (Beckett, 1999). Furthermore, very few empirical studies have focussed
upon risk assessment with juvenile sexual offenders (Prentky et al., 2000). Again, this is
unfortunate, given the incidence of juvenile sexual aggression, and the clear benefits of
distinguishing high- and low-risk adolescent sexual offenders.
Despite these limitations, a number of studies have examined predictive factors of
juvenile sexual recidivism. In keeping with the maxim that past behaviour is the best predictor
of future behaviour, the number of prior sexual offences is perhaps the most robust predictor
of juvenile sexual recidivism (e.g., Rasmussen, 1999; Scram, Milloy, & Rowe, 1991).
A number of studies have found that sexual offence characteristics are predictive of future
offending. Kahn and Chambers (1991) conducted a follow up of 221 juvenile sexual offenders
and found that verbal threats during index offence and denial or victim blame were associated
with sexual offence recidivism in the 7.5% of their sample population who reoffended. Smith
and Monastersky (1986) found that indecent exposure, same age or adult victim, male victim
and stranger-victim were all predictive factors of sexual offence recidivism.
One of the most significant predictors of juvenile sexual offence recidivism is
delinquency. Ageton (1983) identified four variables that could correctly classify 77% of
the juveniles that offended sexually: involvement with delinquent peers, crimes against
persons, attitude towards sexual assault, and family ‘‘normlessness’’. However, a subsequent
discriminant analysis using only delinquency variables revealed that one variable, involvement
with delinquent peers, correctly classified 76% of the cases. Ageton concluded that sexual and
offence characteristics did not significantly improve the prediction model. Delinquency
variables have been found to be predictive of sexual offence recidivism in this population by
others, such as Prentky and Knight (1993) who identified adolescents who continued their
sexual abuse into adulthood as characterized by a history of impulsivity and antisocial
behaviour.
The adult research literature indicates that deviant sexual arousal is one of the better
predictors of sexual offence recidivism (e.g., Rice, Quinsey, & Harris, 1991). However,
evidence for the association between deviant sexual arousal and juvenile sexual recidivism has
thus far been mixed. Kenny, Keogh, and Seidler (2001) found that deviant sexual fantasies
were directly related to juvenile sexual offence recidivism in a sample of 70 male juvenile
sexual offenders. Scram et al. (1991) and Langstrom and Grann (1999) have also reported
that juvenile sexual offence recidivism was significantly more likely to present with deviant
patterns of sexual arousal. However, Kahn and Chambers (1991) did not report significantly
38 T. E. Christodoulides et al.
more deviant arousal in their sample of juvenile sexual offenders. Prentky et al. (2000) suggest
that such studies that utilize ratings of deviant sexual arousal by therapists inevitably have
difficulties in differentiating appropriate and age-inappropriate sexual preferences for juvenile
sexual offenders.
It has been suggested that sexual victimization is predictive of sexual offence recidivism
(e.g., Rogers & Tremaine, 1984). However, evidence to support such a claim with juveniles
has thus far been weak. Rasmussen (1999) found that a history of sexual abuse is only weakly
related to sexual offence recidivism. This finding is supported by others (Knight & Prentky,
1993; Prentky & Knight, 1993).
Most characteristics that significantly distinguish one-off and persistent adolescent sexual
offenders are historical and static (e.g., number of prior sexual offences). Research has not yet
indicated that dynamic factors (factors that are changeable over time, such as victim empathy)
demonstrate strong predictive ability (Hall, 1995; Hanson & Bussiere, 1996). A possible
explanation is that such variables have proven problematic to measure, hence many studies
concentrate upon more easily quantifiable static variables. Of course another possible
explanation is that dynamic factors offer little or no true prediction value of recidivism with
this population.
Two detailed reviews of the adolescent sexual offending risk assessment literature
(Beckett, 1999; Prentky et al., 2000) conclude, in broad agreement with the statement of
National Task Force on Juvenile Sexual Offending (National Adolescent Perpetrator
Network, 1988) that ‘‘currently there are no scientifically validated instruments or criteria
to assess risk of re-offence’’ (p. 29). Therefore, although the literature has produced a number
of studies that have identified factors associated with recidivism in this population, we still
lack risk assessment tools to utiliize when making professional and clinical decisions. What is
needed is a risk assessment measure that is able to distinguish those who will not commit
further offences and those who will; and one that has adequate sensitivity to detect those who
will reoffend and have sufficient specificity not to give false positive predictions of reoffending
in those who do not. The present research aimed to use such criteria in testing the predictive
value of a UK-devised risk assessment tool amongst a sample of juvenile sexual offenders. The
research is, by definition, flawed in that the data was collected as part of normal clinical
practice in an adolescent forensic service, the need to establish tighter definitions of clinician
reported reoffending is important in any consideration as to the generalizability of the current
research and planning of future research.
Method
The sample consisted of 50 adolescents referred to the Northern Forensic Mental Health
Service for Young People prior to the end of 1999. The mean age of the sample was 14.6 years
(standard deviation 2.6).
The Risk Assessment Matrix (Appendix A) comprised Risk Evaluation Stage 1 and Stage
2. The purpose of the matrix as originally devised was to provide an initial set of risk factors
and risk levels, which could be completed by a social worker or probation officer prior to
referral of the case for assessment and possible treatment to the Northern Forensic Mental
Health Service. The second stage of the evaluation was designed to include more detailed
clinical information, which was elicited by a psychologist at the time of the clinical assessment.
The risk factors included were those associated with sexual offending and recidivism in the
available clinical/forensic literature. The Risk Assessment Matrix has been published by the
Derwentside Initiative (Richardson, 1995).
Risk assessment with adolescent sex offenders 39
Stage 1 of the risk assessment included 10 risk factors (scoring range 11 to 33), which
were chosen to be relatively easily identifiable by a social worker or probation officer following
an interview and review of case notes. Stage 2 is made up of 16 risk factors (scoring range 16
to 48), and focuses on areas that have a more subjective component and is to be completed by
clinicians experienced in the assessment of adolescent sex offenders.
Two methods were used to establish reoffending. First, the clinical or forensic
psychologist, who had undertaken the assessment, was asked to indicate whether, to their
knowledge, the individual had committed a further act of sexually abusive behaviour or a
sexual offence, after being assessed or treated within the clinical service. This decision was
based on three main sources of information: social services, youth offending teams, and rereferral
to the clinical service as a result of sexually offending. The majority of the adolescents
were subject to either care and/or supervision orders and consequently there was active
involvement of social services and youth offending teams following the involvement of the
adolescent forensic mental health service. This gave a figure of 23 out of 50 (46%). Secondly,
the Home Office records for reconvictions on the sample were obtained for a minimum period
of two years following assessment up to the end of 2001. This resulted in a figure for
reconvictions of 10 out of 50 (20%).
Results
Clinician reported reoffending
An initial independent samples t test was undertaken to compare the reoffenders with nonreoffenders
on total scores on the overall and two sections of the risk assessment. There were
significant differences between groups on total 1, total 2, and total risk (see Table I). The risk
assessment therefore does have some potential in discriminating the two groups.
However, whilst the risk assessment tool may be demonstrating some statistical validity in
the ability to discriminate risk groups, further analysis is needed to demonstrate potential
clinical utility through predictive reliability.
Risk level 1 evaluation. Logistic regression with replacement of missing data by median using
all the data points in Section 1 gives a successful prediction rate of 81.6%, with a 54.0%
successful prediction by chance. Forward conditional logistic regression with normal exit and
enter parameters (0.15 and 0.05) indicated two significant statistical predictors in this group
of variables, these being Family Adjustment and Number of Victims, giving a successful
prediction of 72.0%.
It is important in a risk assessment to achieve adequate sensitivity (i.e., few false
negatives, whilst maintaining false positives at a low enough level for the measure to have
some clinical utility as a screening instrument). This was best achieved using all the data
points with a cut-off score of 22 or less (actual range 13_/31) where there is 86.9% sensitivity,
Table I. Mean, standard deviation, t, and p values for groups on risk assessment tool
Reoffended
(n_/23)
No reoffence
(n_/27) t p
Stage 1 24.66 (3.97) 19.91 (4.39) 3.97 B/0.01
Stage 2 33.89 (4.43) 27.35 (6.62) 4.10 B/0.01
Total 58.55 (7.49) 47.26 (10.45) 4.30 B/0.01
40 T. E. Christodoulides et al.
with a specificity of 70.3%. The Kappa value or prediction above chance is 0.56, indicating
that using this cut off produces a 56.4% improvement above chance. Using simply the
identified variables of Family Adjustment and Number of Victims scores (actual range 2_/6) to
produce a cut off, it was not possible to achieve an adequate sensitivity (greater than 70%)
without the false positive rate rising above 50%.
Risk level 2 evaluation. Logistic regression using all the data points in Section 2 gives a
successful prediction rate of 82.0%, with a 54.0% successful prediction by chance. Stepwise
linear regression with normal exit and enter parameters (0.15 and 0.05) indicated two
individually significant statistical predictors in this group of variables, these being Social
Competency and Family History of Offending, giving a successful prediction rate as good as
including all the variables at 82.0%.
Adequate sensitivity was best achieved using a cut-off score of 30 or more (actual range
17_/45), where there is 86.9% sensitivity, with a specificity of 66.7%. The Kappa value was
0.53. Using just ‘‘Social Competency’’ and ‘‘Family History of Offending’’ and a cut-off score
of 4 or more, it was possible to obtain 86.9% sensitivity with a specificity of 74.1%, with a
Kappa value of 0.56.
Total risk evaluation. Logistic regression using all the data points in Stage 1 and Stage 2 gives a
successful prediction rate of 100.0 with a 54.0% successful prediction by chance. This is
simply a statistical anomaly as the number of variables is over half the number of subjects.
Stepwise linear regression with normal exit and enter parameters (0.15 and 0.05) indicated
two individually significant statistical predictors in this group of variables, these again being
Social Competency and Family History of Criminal Activity giving a successful prediction
rate as good as including all the variables at 82.0%.
Combining the 4 variables: Family Adjustment and Number of Victims from Stage 1,
and Social Competency and Family History of Offending from stage 2, a reasonable model in
terms of sensitivity and selectivity was possible using a cut off score of 8 or more. This model
has a specificity of 86.9% and a sensitivity of 62.3%, and a Kappa value of 0.49.
Reconviction data
Risk level 1 evaluation. As only 10 of the sample had reconvicted by home office data, a logistic
regression using all variables was not appropriate, as number of variables exceeded the sample
size. Forward conditional logistic regression with normal exit and enter parameters (0.15 and
0.05) indicated two significant statistical predictors in this group of variables, these being
Location and Age of Victims giving a successful prediction of 86.0%. However, no
appropriate cut off score could be determined with adequate sensitivity with all the
reconvicted sample and 31 of the non-reconvicted sample obtaining scores of 4 or more.
Risk level 2 evaluation. Forward conditional logistic regression with normal exit and enter
parameters (0.15 and 0.05) indicated one significant statistical predictor in this group of
variables, this being Previous Convictions giving a successful prediction of 86.0%. However,
no appropriate cut-off score could be determined with adequate sensitivity with 5 of the
reconvicted sample and 35 of the non-reconvicted sample having no previous convictions
prior to the proceedings leading to the referral which prompted the risk assessment.
Risk assessment with adolescent sex offenders 41
Total risk evaluation. Combining the three variables: Location, Age of Victims (from stage 1),
and Previous Convictions (from Stage 2) of the risk assessment tool, again did not result in the
ability to obtain a cut-off score with adequate sensitivity or specificity.
Discussion
As expected, clinicians in the present study reported far higher rates of recidivism than official
recidivism data from the Home Office. It is a consistent finding in the juvenile (and adult)
sexual offender literature that officially recorded recidivism is a significant underestimation of
the real number of offences committed (e.g., Hanson & Bussiere, 1998). This is believed to be
due to a number of factors, not least in the UK the reluctance of the Crown Prosecution
Service to rely on child witness testimony, where it may be considered to be unreliable. In
addition the victims’ families may prefer that their child does not experience the potential
distress of providing evidence.
It is not uncommon for such low rates of official recidivism to produce inconclusive
findings in studies attempting to distinguish between juvenile sexual recidivists and nonrecidivists
(Kahn & Chambers, 1991; Rasmussen, 1999). The present research was no
exception; the risk matrix was unable to distinguish recidivists and non-recidivists on the basis
of official reconviction data. However, it did successfully distinguish between juvenile sexual
recidivists and non-recidivists using clinician’s judgement of recidivism as outcome. The four
most predictive variables (Number of Victims, Family Adjustment, Family History of
Offending, and Social Competency) are consistent with prior research on juvenile sexual
offending recidivism. Number of prior victims has been associated with recidivism
(Rasmussen, 1999), as have familial variables (Ageton, 1983) and factors related to social
skills deficits (Katz, 1990; Kobayashi, Sales, Becker, & Figueredo, 1995; Marshall, 1993).
The study does not have sufficient numbers, nor are the referrals from a sufficient range
of sources, to conclude that a specific group of factors can be reliably used to predict
recidivism. For example, there was no marked difference in the predictive validity of Stage 1
and Stage 2 of the risk assessment. Each of the statistical models produced a moderate to
excellent level of sensitivity (62.3_/86.9%), and moderate to excellent specificity (66.7_/
86.9%). The Kappa values of all the statistical models were approximately 0.5. Rather than
suggesting that there is a specific set of factors, it appears that it is the process of utilizing a
structured assessment protocol comprising a set of relevant factors, which is helpful in
predicting recidivism. The discovery of one or more ‘‘magic’’ factors is not likely, but the value
of objective review of criteria highlighted in the sexual offender literature is clearly indicated.
This seems a reasonable account for the number of previous findings on juvenile sexual
offending that have produced similar predictive results, although they measured different
factors (Kenny et al., 2001).
The use of clinician’s judgement of recidivism (as opposed to official statistics) has
produced a potentially valuable risk assessment tool in this case, but has also resulted in
methodological problems, in particular inter-rater reliability. In the present study, clinicians
were not provided with strict criteria for distinguishing recidivists versus non-recidivists. This
represents a core methodological flaw to the study. For example, one clinician may have
judged that registration on the child protection register for further alleged incidents of sexual
abuse signifies recidivism, whereas other clinicians may have utilized less stringent criteria.
One way to address this would be to provide clinicians with specific guidelines with which to
judge reoccurrance of abusive behaviour. This might include a range of criteria from the
almost certain (e.g., police report of reoffending, but decision not to prosecute), through to
42 T. E. Christodoulides et al.
hearsay evidence (e.g., uncorroborated report from relative), and a clear demarcation of what
would constitue acceptable evidence. Another possibility may be to seek judgements from an
independent rater; decisions could be made on basis of multiple sources of information
including: clinical case notes, interviews with the adolescent and lead clinician, and survey
information from relevant statutory bodies (e.g., police, social services, and youth offending).
Although the specificity and sensitivity ratings achieved within the present study are
significantly above chance, there is still a large part of the variance unaccounted for. This is
somewhat inevitable given the complex task of risk assessment. There are unquestionably
factors that will have contributed to the variance that are unmeasured here. One such factor
may be ‘‘opportunity to reoffend’’. For example, an incarcerated juvenile sexual offender, who
is systematically observed, has their freedom restricted, and is kept isolated from potential
victims, is less at risk of reoffending simply by virtue of restricted opportunity, regardless of
their presenting a number of historical and/or dynamic risk factors. Kenny et al. (2001)
suggest that opportunity for reoffending is extremely problematic to measure, but very likely
to affect recidivism outcome.
Recidivism rates are notoriously sensitive to the populations under study (Hanson &
Bussiere, 1998). In the present study, the adolescent sample referred for assessment and
treatment at a forensic outpatient mental health service may not be representative of the
juvenile sexual offender population as a whole. Clearly there is a need to replicate the study to
establish the reliability of the tool. One group to study might be those referred to Youth
Offending Teams but not referred for assessment within adolescent mental health services.
Nevertheless, this tool has demonstrated its utility in distinguishing recidivists from nonrecidivists
in a sample of juvenile sexual offenders. As such, it has potential value in informing
professionals’ decisions which should be based upon level of presenting risk. However, it also
has further potential in assisting clinicians in the identification and modification of contextual
and clinical risk factors, and in developing and implementing a risk management plan. This is
consistent with a reconceptualization of the concept of risk that Sheldrick (1999) and
Steadman et al. (1993) have called for, encouraging clinicians to move away from a focus
upon dichotomous decisions (‘‘will they reoffend?’’) to an emphasis upon ongoing management
and treatment issues. This allows the clinician to concentrate upon potential for change
and takes into account the dynamic nature of risk. This is perhaps even more applicable to
adolescents, given their potential for growth during such a period of transition.


Test
Table of Contents
Top