Sixty-nine teenage child molesters received a 3-month regimen of vicarious sensitization (VS) within the context of a randomized wait-list control group design. An adjunct to specialized cognitive therapy, VS is a form of aversive conditioning the aim of which is to decrease sexual arousal to prepubescent children. Perpetrators were alternately exposed to an audiotaped crime scenario designed to evoke deviant arousal followed immediately by an aversive video vignette. The aversive stimuli portray adolescent sex offenders contending with negative social, emotional, physical, and legal consequences of their sex crimes. Subjects received approximately 300 VS trials over 25 sessions. Results based on both phallometric data and self-report measures showed significant decreases in deviant arousal for youths who received VS. Wait-listed youth did not, despite continuing in weekly peer-group therapy. When VS was later administered to wait- listed youths, they too showed a significant treatment effect. Erectile responses to adolescent girls were unaffected by the procedure. Three-month followup data indicated that treatment gains were maintained.

Introduction

Studies of adult sex offenders show that many, perhaps the majority, began committing sex crimes in their teenage years or earlier (Abel, Mittelman, & Becker, 1985; Groth, Longo, & McFadin, 1982; Longo & McFadin, 1981). Roughly 25% of sex abuse arrests (U.S. Department of Justice, 1993) and about 20% of reported rape victimizations (U.S. Department of Justice, 1979-1992) involve perpetrators under age 21. Based on youth self-report, it is estimated that each year between 195,000 and 450,000 American youths commit sexual assaults involving force (Ageton, 1983). This figure does not include child molestation, the most common juvenile sex offense (Ryan & Lane, 1991). Conservatively, at least a half-million juveniles commit a hands-on sex crime every year (Weinrott, 1996).

One response to the proliferation of juvenile sex offenders (JSO's) has been the advent of specialized treatment programs. As recently as 1980, there was only one such program. According to a recent survey, there are now nearly 1000 operating in North America (Freeman-Longo, Bird, Stevenson, & Fiske, 1995). Programs are predicated on the notion that the basic imprint of sexual orientation occurs in adolescence (Becker, Cunningham-Rathner, & Kaplan, 1986). The learning process is thought to involve the interaction of three basic factors: a) social influences, primarily through peers and immediate family, b) observed and firsthand sexual experiences that affect the content of erotic fantasies which are c) reinforced and further shaped by masturbatory practice (Laws & Marshall, 1990). For example, a child molester may himself have been molested or observed sexual exploitation and learned that certain forms of deviant sexual expression are reinforcing. These events form the content of his sexual fantasies which shape both his masturbation and sexual advances. High levels of deviant arousal have been associated with prior sexual aggression (Malamuth, 1986; Malamuth, Heavey, & Linz, 1993; Quinsey & Earls, 1990) and recidivism (Malcolm, Andrews, & Quinsey, 1993; Quinsey, Rice, & Harris, 1995; Schram, Milloy, & Rowe, 1991; Serin, Malcolm, Kahanna, & Barbaree, 1994).

The implications for a prevention or an early intervention program are straightforward: (a) alter distorted thinking about human sexuality that emerged through interaction with or observation of peers and family members, (b) prepare one for age-appropriate, consensual relationships, and (c) reduce deviant sexual fantasies and the masturbatory practices that both reinforce and are evoked by these fantasies (Salter, 1988). Many treatment programs have all three objectives, but those for JSOs typically avoid direct reduction of deviant sexual arousal except by means of thought stoppage or appropriate fantasy substitution. The focus is almost exclusively on cognitive restructuring (Davis & Leitenberg, 1987; Lakey, 1994; National Council on Juvenile & Family Court Judges, 1993). Eschewing aversive conditioning techniques may be a serious omission, particularly in light of studies by Marshall and Barbaree (1988a, 1988b). These investigators found that adult sex offenders who receive treatment that included direct methods to reduce deviant arousal were less likely to reoffend than a group of men whose otherwise similar treatment lacked such a component. Moreover, Rouleau, Abel, Mittelman, Becker, and Cunningham-Rathner (1986) found evidence that such a component needs to be introduced at or near the beginning of a treatment program, otherwise sexual aggressors tend to harbor deviant fantasies throughout the later stages of therapy and are more likely to drop out prematurely.

Nearly all conditioning methods used with sex offenders rely on scripts, audiotapes, or free fantasy to induce deviant arousal, that is, arousal to an illicit sex act. Often an audiotaped version of the offender's crime(s) is used for this purpose. At some point in the procedure, either an aversive stimulus is introduced or the crime scenario (e.g., rape, child molestation, exposing) begins to take on an aversive quality through repetition. Virtually all aversive procedures require dozens, if not hundreds, of trials over a minimum period of 10 days to 12 weeks. Treatment of six months or more is not unusual.

Fewer than half the programs for JSOs rely on some form of aversive conditioning or satiation (Freeman-Longo et al., 1995). The most popular techniques and the percentage of JSO programs utilizing each is as follows: covert sensitization 36%, verbal satiation 18%, masturbatory satiation 17%, olfactory conditioning 11%, and faradic conditioning 3%. Unfortunately covert sensitization may be ineffective with JSOs (Emerick, in press), and verbal satiation reduces deviant arousal very slowly, and not at all in younger adolescents (Hunter & Santos, 1990). There are political objections to the use of masturbatory satiation with juveniles, and (unfounded) ethical concerns that preclude olfactory and faradic conditioning in most clinical settings.

So, even though over 90% of JSO treatment programs endorse the objective of reducing deviant urges, fantasies and masturbatory practices (Knopp, Rosenberg, & Stevenson, 1986), there is no readily available means to accomplish this. What is needed is a conditioning procedure that is politically palatable, cost-effective, easy to disseminate, and readily acceptable to clients. The present investigation represents the first experimental test of vicarious sensitization (VS), a new alternative for JSOs.

Methods

Subjects and Screening

Subjects were recruited from outpatient JSO treatment programs, private practitioners, and probation officers in the Portland, Oregon metropolitan area as well as from Echo Glen Children's Center, a Washington State institution near Seattle. To be considered for VS a youth had to: a) be male, b) be age 13-18 at the time of referral, c) have committed a hands-on sex offense against a child at least four years younger, d) admit having done so, e) volunteer for VS to reduce arousal to children, and f) have at least six months remaining in his core treatment. Written consent was obtained from 118 boys and their guardians. One hundred and eight of these youths were assessed phallometrically (see below), 15 of whom were excluded due to low overall or deviant arousal.1 Twenty-four youths withdrew from the study prior to completion.2 The final sample consisted of 69 boys who fulfilled all assessment and treatment obligations over a six-month period. Table 1 contains demographic and sexual history data. It is worth noting that virtually all youths had been adjudicated and were participating in specialized sex offender treatment at the time of referral. Most treatment programs utilized a peer-group format supplemented by individual and/or family therapy. Typical treatment activities include accepting personal responsibility, cycle identification, empathy training, anger management, elimination of "thinking errors," social skills training and relapse prevention. These components have been described elsewhere (Emerick, in press; Kahn, 1990; Ryan & Lane, 1991; Steen, 1993). All youths continued in their core treatment while participating in the present study. Very little information about a youth's performance in VS was conveyed to referring therapists until participation ended. A federal Certificate of Confidentiality was obtained to preclude unauthorized access to data.

Design and Setting

Following the initial assessment (see below), qualifying youths were randomly assigned to either an immediate vicarious sensitization group (IT) or to a three- month waiting list (WL) condition. T-tests and chi-square tests of variables presented in Table 1 yielded no significant pre-existing differences between the two groups. Thirty-five youths in the IT group received 25 sessions of VS twice per week, after which they were reevaluated. Their 34 wait list counterparts were also reassessed prior to receiving the identical three-month regimen of VS. Three months after the second assessment, subjects in both conditions were again reassessed. All assessment and treatment sessions were individual and took place in a mobile laboratory (for outpatients) or in the sexual laboratory at Echo Glen. Both labs were identical, consisting of a private viewing room equipped with a 27" color monitor, headphones, wall-mounted video camera to ensure attending, and electronic light bar for rating sexual arousal. An adjoining control room contained a Medical Monitoring Systems RS3000 phallometric recorder driven by an IBM compatible PC, Super VHS video cassette recorder, an audio cassette deck, and a small video monitor projecting a head shot of the subject. Erectile responses were transmitted by means of D. M. Davis mercury strain gauges. Both labs were also equipped with two-way intercom.

Production of Aversive Stimulus Materials

Thirty-one aversive vignettes were videotaped. Scripts were written by the investigators and actors were cast as perpetrators, peers, teachers, parents, police, and so forth. Ranging in length from two to eight minutes, tapes depicted a wide variety of legal, social, physical, and emotional consequences experienced by juvenile sex offenders. Examples of vignettes are: (a) JSO caught in the act by victim's mother; (b) JSO's girlfriend learns of his offending and reacts with disgust; and (c) angry victim who has grown up confronts JSO. Ten vignettes were available in four versions: younger brother, younger sister, little boy, and little girl. Twelve vignettes apply to both incestuous and non-incestuous situations so only two versions were needed, one for male victims and one for their female counterparts. Finally, nine stories accommodated all perpetrators regardless of the victim's sex or relationship to the offender. The different versions permitted matching of video scenarios to a youth's victim(s).

Crime Scenario

The crime scenario was a composite description of the youth's offense(s) and child-oriented sexual fantasies. Scenarios were developed during a single interview with each youth. Offense descriptions focused heavily on setting factors, physical characteristics of the victim, emotional states, grooming, self-talk, justifications, and intent. The initial crime scenario for a youth stopped just short of actual molestation with more sexual content added in later sessions. Length of a typical fantasy scenario was 1« minutes. Youths with victims of both sexes had two crime scenarios which were alternated on a session-by-session basis.

Aversive Vignettes

After listening to the entire crime scenario on audiotape, a portion of an aversive video was presented. Video vignettes were divided into segments, as few as two and as many as seven. Segments ranged in length from 35 seconds to 3« minutes. Each pairing of arousing and aversive stimuli constituted a single VS trial. While the deviant scenario remained the same across trials (until revised), the aversive videotapes were shown in sequence. Youths circulated through the series of video vignettes three times for a maximum of 333 trials.3 Youths were exposed to a total of 275-300 trials or 11-15 per session. After a subject reported no perceptible arousal for any trial occurring in three consecutive sessions or in any four of five consecutive sessions, more sexual content was added to the crime scenario.4

Youths were actively discouraged from masturbating to child-oriented sexual fantasies. To help prevent this, JSOs were provided with a supply of wallet cards, each of which contained a brief description of all 31 aversive vignettes. Subjects were instructed to refer to the wallet card when experiencing deviant urges or fantasies outside the lab and to envision one of the most anxiety producing vignettes until the urge subsided. Youths were also encouraged to leave the setting in which the deviant urge or fantasy occurred.

Assessment

Three separate sets of phallometric stimuli were utilized because pilot testing had shown low correlations between responses to audio stimuli depicting sexual scenarios and commercially produced slides of inactive nudes. And because of mounting ethical concerns about the use of child slides, there was a growing need to develop visual stimuli that were less objectionable. Since evidence of convergent validity is lacking (at least with respect to juveniles), use of multiple measures was essential. During each assessment, the video and audio stimuli were presented in a single session with the slide portion following within 48 hours. Strain gauges were calibrated at the beginning of each session to ensure linearity.

Video Phallometric Measures

A series of 25 2«-minute color videotapes was produced. These depict males and females belonging to different racial and age groups ranging from 4-16 years old. Each tape portrays a single actor in a swimsuit, who sits at the edge of a pool, dries him/herself, eats a piece of fruit, and walks along the pool deck. There are occasional close ups of genital areas and breasts, but no nudity or sex acts. The sound track consists solely of background music. In addition, there was a neutral tape containing a fishing scene. From still photos, subjects selected the most attractive actor in each of eight age-by-sex categories. Subjects were asked to narrate the action to ensure their attention. The videotapes of these actors, plus the neutral tape, were shown in random order (by subject) and assessment.

Audio Phallometric Measures

From the set of audio cues produced by Becker and Kaplan (1988), seven were selected for use in the present study. These 2-minute graphic descriptions of sexual encounters included coercive sex with a girl age 6 or 7, coercive sex with a boy age 6 or 7, forced sex with a girl age 10 or 11, forced sex with a boy age 10 or 11, consensual sex with a teenage girl, and consensual sex with a teenage boy. In addition there was a neutral (nonsexual) description of a party. The seven audiotapes were presented in random order (by subject) and assessment.

Slide Phallometric Measures

Twenty-four slides of inactive nudes were extracted from the national standardization series produced by Farrall Instruments Company and evaluated by Laws, Gulayets, and Frenzel (1995). These were transferred to videotape to allow intermittent placement of a red dot in one corner of one of the slides for three seconds. To maximize attending, subjects were required to signal verbally when a dot appeared on the monitor. There were four slides each of males and females in three age categories: 3-6, 7-9, and 15-17. These were supplemented by four neutral (landscape) slides. Exposure time was 2 minutes per stimulus. The order of presentation was identical for all subjects and assessments. This was determined by randomly assigning one stimulus from each category to a block, and then randomly ordering stimuli within blocks.

Adolescent Sexual Interest Cardsort (Becker & Kaplan, 1988)

Converted to a questionnaire, this 64-item instrument requires youths to rate (on a 5-point bipolar scale) their subjective arousal to a wide variety of short sexual scenarios. Sixteen rationally derived subscales have been formed (Hunter, 1991). Since vicarious sensitization was developed to reduce sexual arousal to children, three scales were of particular interest: nonaggressive sex young female, aggressive sex young female, and aggressive sex young male. Following are sample items, one from each of these scales:

1. My penis is moving in and out of a five year old girl's mouth. It really feels good.
2. I am pushing my finger between the legs of a 6 year old girl. I am too strong; she can't get away from me.
3. I've pulled a 7 year old boy into my bed. I'm rubbing my penis against his butt.

Self-Perception Profile for Adolescents (Harter, 1988)

This 45-item self-esteem inventory contains nine subscales tapping eight specific domains as well as global self-worth. Using a two-stage, forced choice format, youths rate the degree to which various physical, academic, behavioral, and social characteristics match their self-perceptions. Two subscales were of particular interest: behavioral conduct and global self-worth, since these were apt to be the most sensitive to either positive or adverse effects of VS.

Social Validation and Clinical Significance

Youths and their parents rated VS on dimensions of effectiveness, humaneness, and overall satisfaction. Rates of deviant urges, fantasies, and masturbatory practices were also obtained from each subject at each assessment.

Results

Phallometric Measures

Erectile responses to each phallometric stimulus were scored in two ways: (a) maximum percentage of a full erection, and (b) latencies (in seconds) to 25% of a full erection. All scores pertained to the interval from stimulus onset to offset. The two sets of scores yielded results that were consistent with one another. Simplicity, convention, and psychometric integrity dictate that results be presented in terms of percentage of full erection (Barbaree & Mewhort, 1994). Only where elaboration is needed are latencies introduced.

After an examination of individual arousal profiles, it was decided to report scores for all subjects on both female and male stimuli. Unlike adult pedophiles whose measured sexual preferences normally correspond to the age and sex of their victims, almost all JSOs responded more to female than male stimuli. With only two exceptions, subjects with exclusively male child victims showed higher levels of arousal to females. Although their responses to male stimuli were clearly higher than those for youths who had only female victims, it seems that JSOs who abuse males often do so more on the basis of opportunity than sexual preference. Partitioning the sample by sex of victim would have resulted in statistical tests of insufficient power.

Two sets of analyses were performed on all phallometric measures: MANOVAs using each age-by-sex stimulus category as a dependent variable, and ANOVAs on composite measures (i.e., means) across pre-pubescent age categories for each sex. Alpha was set at 0.1 for omnibus MANOVAs, rather than at the conventional .05 level for two related reasons: tests of the interaction effect lacked power, and a Type II error should not be tolerated when evaluating a new method in a young field using imperfect criterion measures.6 Composite scores appear in Table 2 along with those corresponding to female and male teens. On the few occasions when pre-existing (Assessment I) differences emerged from t-tests then a MANCOVA or ANCOVA was performed using the score from Assessment I as a covariate. Planned orthogonal comparisons were performed both at the individual category level and on composites. The relevant comparisons were a) between the immediate treatment (IT) and wait list (WL) groups at Assessment II, b) within the IT group comparing pretreatment (Assessment I) scores to those from each of the later two assessments, and c) within the WL group comparing results from Assessments II and III. When a MANCOVA or ANCOVA was performed, between group comparisons were conducted using adjusted means. Table 3 summarizes tests of specific hypotheses on each phallometric stimulus category.

Video Phallometric Measures

Figure 1 shows the maximum percentage of a full erection (by Assessment) averaged over the three female child age categories. JSOs who received vicarious sensitization immediately after referral decreased their arousal from a mean of 38% to 23% of a full erection. Three months later, the mean had fallen to 20%. Those on the waiting list showed no change in deviant arousal from their pretreatment level of 45% before falling to 31% after receiving VS.

A two group by three Assessment MANOVA was performed using the three female child category scores as dependent variables. A significant main effect was obtained for the assessment (time) factor (Wilks F(6,62) =.69, p<.001). Both the group effect (p<.08) and interaction (p<.09) were significant. As seen in Table 3, all of the relevant contrasts were also significant. Parenthetically, post hoc Duncan Multiple Range Tests yielded a significant group difference at Assessment III, favoring those who received VS immediately. This was true for the 9-10 year old category (p<.03) and the composite of all three video child measures (p<.03). It is noteworthy that sexual arousal to female peers remained relatively high and quite stable for both groups throughout the course of the study. Arousal to male peers declined.

Initial levels of arousal to the male videos were much lower (see Figure 1) with a composite score of only 17% for the IT group and 21% for those on the waiting list. Still these levels dropped significantly during the course of the study. With the three male child category scores serving as dependent variables, a MANOVA generated a significant main effect for assessment, F(6,62)=.77, p<.01. Neither the group factor nor the interaction were significant.

Audio Phallometric Measures

The most striking result emerging from this stimulus set is the high level of arousal elicited by both female and male sexual scenarios. At Assessment I, the composite female child mean was a hefty 90% for JSOs in the IT group and 92% for those on the waiting list.7 As shown in Figure 1, subjects who received VS immediately, decreased their arousal to a mean of 63% at Assessment II, a decline that continued into the follow-up phase (55%). JSOs on the waiting list showed a non-significant decrease of 8% from Assessment I to II. The MANOVA on the two female child audio variables yielded a significant main effect for the assessment factor, F(4,64)=.52, p<.001, and significant results for both the group factor (p<.1) and the interaction term (p<.1). Planned comparisons showed that at Assessment II, IT youths significantly outperformed their untreated counterparts on both female child categories and their composite (see Table 3). When boys on the waiting list were eventually exposed to VS, they too showed a sizeable reduction in arousal for both child categories and on the composite. Neither group demonstrated a reliable decrease in arousal to the age-equivalent consensual scenario.

Nearly half of the sample showed a post-VS maximum below 30% of a full erection to both of the female child audio scenarios. But because the post-treatment mean levels remained fairly high, a secondary analysis of latencies was performed to determine whether the rate of arousal had been differentially and substantially affected. Using a composite of both female child audios, Figure 2 shows the number of seconds elapsed between stimulus onset and obtaining 25% full erection, a mildly perceptible level of arousal. The IT group mean of only 43 (of a possible 120) seconds at Assessment I rose to 75 seconds following VS and remained stable thereafter. JSOs in the WL group showed no discernable change prior to receiving treatment but a similarly dramatic decrease in rate thereafter. The MANOVA on latency scores yielded significant main effects for group, F(2,66)=.89, p<.03, and assessment, F(4,64)=.37, p.<.001, as well as for their interaction, F(4,64)=.79, p<.004. Planned comparisons favored the IT group at Assessment II for both dependent variables (p<.001). The relevant within group comparisons were also highly significant. With respect to graphic sexual stimuli, many youths continued to respond to child content. However, it appears that VS reduced the rate at which this occurs. The clinical implications are very positive for clients who have learned to invoke a coping response when confronted with a high risk situation.

The male audios generated a similar set of results, although both the initial and final arousal percentages were lower than on female stimuli. IT youths dropped from 63% of a full erection to 44% after VS and maintained (at 41%) thereafter. From an initial mean level of 71%, boys in the WL condition showed a non-significant decrease of 3% at Assessment II and then dropped to 47% following VS (see Figure 1). The MANOVA yielded a significant main effect for assessment, F(4,64)=.58, p<.001, and the interaction, F(4,64)=.87, p<.07. Reliable differences between groups at Assessment II and within groups pre-post were obtained on both child male age categories and their composite (see Table 3).

Analysis of latencies (to 25% of a full erection) were identical to those conducted on the female audio scores both in form and outcome. Assessment I means were 72 and 67 seconds respectively for the IT and WL groups on the composite measure. The IT mean increased to 94 seconds after VS while remaining essentially unchanged at Assessment II for those on the waiting list. Treatment gains were maintained for IT youths (96 seconds) at Assessment III and were replicated in the WL group following VS (90 seconds). Results of the MANOVA showed a main effect for assessment, F(4,64)=.57, p<.001, and the interaction, F(4,64)=.88, p<.08. The four planned comparisons between and within groups were all significant for the composite measure and the component category scores. Again, VS lowered the rate of deviant arousal even when the absolute level achieved remained (for some youths) problematic. It is noteworthy that latencies to both the teenage boy and girl audios remained stable over time.

Slide Phallometric Measures

The pattern of change based on responses to female slides was similar to that derived from the video and audio tapes. On a composite measure of eight female child slides the initial level of arousal was 47% for IT youths and 55% for their WL counterparts. This difference was not statistically significant. Following VS, the IT mean fell to 23% and continued dropping to 18% during the next three months. JSOs on the waiting list showed a nonsignificant decrease of about 10% at Assessment II followed by an additional 17% reduction (to 28%) following VS. The MANOVA of the two age category means yielded significant main effects for group, F(2,66)=.87, p<.01, and assessment, F(4,64)=.33, p<.001. The interaction term was also significant, F(4,64)=.85, p<.05. Planned comparisons showed a significant advantage for IT youths at Assessment II on the composite measure (p<.001) as well as on each of the category scores (p<.001). The relevant within group comparisons were also significant (see Table 3). Post hoc Duncan Multiple Range Tests comparing the two groups at Assessment III favored IT youths on the composite measure (p<.05) and specifically in the 3-6 year old age category (p<.01). However if one adjusts scores at Assessment II and III on the basis of (non-significant) pre-treatment differences, then the disparity at Assessment III more or less disappears.

Using male child slides as the criterion, initial levels of deviant arousal were low and continued to drop during all phases for both groups. This pattern was similar to that obtained on male video tapes and it suggests a crystallizing of heterosexuality over time. A MANCOVA using the pre-treatment scores as covariates generated a significant main effect for assessment, F(2,64)=.83, p<.01, but not for the interaction. Hence, the impact of VS was considered negligible.

Adolescent Sexual Interest Cardsort

Because phallometric measures were utilized for purposes of subject selection, every youth demonstrated moderate to high arousal to at least one child-related stimulus (and invariably more). In other words, non-responders simply did not exist in the current sample. However, on a self-report measure like the cardsort, denial and minimization were fairly prevalent. Indeed the group means at Assessment I did not exceed 0.62 for either group on any of the relevant scales: nonaggressive sex with young female (NSYF), aggressive sex with young female (ASYF), and aggressive sex with young male (ASYM). Still, many youths for whom Assessment I scale scores were positive but near zero did rate one or two items as arousing. To examine the effects of VS on those boys who reported deviant arousal at Assessment I, two steps were taken. First, a youth was eliminated from the analysis if his highest rating to every item in a given scale was zero or below. This disqualified twelve subjects from the analysis of NSYF, 15 from ASYF, and 38 from ASYM. Second, instead of using scale scores, the value of the highest rated item was substituted as the dependent variable in a series of three ANOVAs. Table 4 shows the sample size, group means, and standard deviations at each assessment. For NSYF, the highest rating by IT youths fell from a mean of 1.53 to -.06 following VS. Three months later it continued to hover around zero. From Assessment I to II, youths in the WL condition showed a decline less than half the magnitude of their IT counterparts. Following VS, boys in the WL group also dropped to just below zero. The ANOVA yielded a significant main effect for assessment, F(2,110)=40.25, p<.001, and the group-by- assessment interaction (F (2,110)=5.38 p<.01). Planned comparisons between groups at Assessment II (p<.05) and within groups prior to and following VS (p<.01) were all significant. It is noteworthy that post hoc tests showed that JSOs on the waiting list showed a significant decrease in their ratings during the three months they were awaiting VS (p<.01).

The results on ASYF were virtually identical (see Table 4). Again the ANOVA showed a significant main effect for the assessment factor, F(2,104)=42.83, p<.001, and the interaction, F(2,104)=6.33, p<.01. The group comparison at Assessment II was highly significant (p<.01) as were all the relevant within group planned comparisons. The only notable difference from the previous set of findings was that the drop in ratings that occurred among JSOs while on the waiting list was not statistically reliable.

Analysis of ASYM ratings was hampered by low power due to subject loss. Boys in the immediate treatment group showed a rating reduction of nearly 1-1/2 scale points (to below zero) following VS. Their rating at Assessment III was even lower. Boys on the waiting list showed a fairly substantial decrease prior to treatment and then an even larger drop there-after. Both groups highest rated item at Assessment III was below zero. The ANOVA yielded a significant main effect for assessment, F(2,58)=18.99, p<.001, but not the interaction.

Self-Perception Profile

Both groups began the study approximately one standard deviation below the norms reported by Harter (1988). By the end of the study JSOs in both groups were well within the normal range on both the behavioral conduct and global self-worth scales. There was no empirical support whatsoever for the notion that VS had an adverse effect on self-esteem. Virtually identical at Assessment I, there was a slight (nonsignificant) difference between groups at Assessment II, which actually favored the boys who had been exposed to VS. Interestingly, wait listed youths evinced no improvement until after receiving VS, this despite their ongoing participation in peer-group therapy. The pattern of ratings was similar on both scales and the results of a MANOVA and univariate ANOVAs showed a significant increase in self- esteem over time across groups. Post hoc tests revealed no differences between the two groups at any point.

Social Validation and Clinical Significance

Consumer ratings have been collapsed across groups, since there were no differences between them on post-VS ratings nor was there a significant change during the three month follow-up for the IT boys. On the item "Since you have received VS, what has happened to your deviant arousal outside the lab?", 80% reported that "It's gone way down" and another 14% reported it "having gone down some." No one reported an increase in deviant arousal. Sixty-five percent of youths reported masturbating to deviant fantasies during the two weeks prior to VS compared to only 17% during the two weeks following completion of treatment. The estimates for deviant urges and fantasies were similar. Seventy-eight percent of JSOs rated VS as equal to or even more helpful than their core therapy. Sixty-two percent agreed that VS improved their attitude in group or individual therapy. Eighty-nine percent claimed that VS helped them understand how most people feel about sex offenders and 87% conceded that VS helped them realize that their sexual problems were more serious than they had originally thought.

Parent ratings were obtained from 40 of 48 (83%) of those JSOs residing in the community. Nearly half (48%) felt that VS greatly improved their son's attitude toward core therapy and another 25% believed it helped "a little." Seventy-five percent felt that they would definitely recommend VS to other parents of JSOs, 7% would with reservations and 15% couldn't say. Only one parent said s/he could not recommend the procedure.8 Eighty percent felt that VS definitely, or probably, made their sons take sex abuse more seriously and 87% felt that VS was worth the time, effort, and inconvenience. Most gratifying, 62%, believed their sons felt much more normal regarding their sexual arousal pattern with another 32% claiming a positive, but less dramatic, change. Nearly all respondents thought that their sons considered VS either extremely helpful (62%) or somewhat helpful (33%) in preventing a reoffense.

Discussion

On all phallometric composites and both relevant cardsort scales, vicarious sensitization reduced arousal to prepubescent girls. Youths who had not yet been exposed to the procedure showed no consistent drop below baseline levels until they, too, received VS. The decreases in measured and reported arousal applied solely to young girls, not their teenage counterparts. Changes in homosexual arousal were more difficult to interpret, owing to generally low baseline levels on phallometric and self-report measures, except those derived from graphic audiotapes. While arousal to young males decreased on all measures, only those changes on the audiotapes could be attributed to VS. Overall, the effects of VS appear to be much stronger than those obtained using covert sensitization (Emerick, in press) and faster than those achieved by verbal satiation (Hunter and Goodwin, 1992), the two most popular aversive procedures.

The clinical relevance of these findings can be evaluated in two ways. First, youth responses on the cardsort combined with youth and parent consumer ratings indicate that sexual interest in children was greatly diminished following VS. That youths reported less deviant masturbation and more normal sexual impulses bodes well for the social value of this procedure. Second, sexual arousal to children has been associated with recidivism among JSOs in both prospective (O'Shaughnessy, McBride, Gretton, & Hare, in preparation) and retrospective research (Schram et al., 1991). While the present study was not an experimental test of this relationship, the extent to which the effects of VS are durable should reduce reoffending in the same way that a method that reduces smoking should prevent lung cancer.

Despite advantages over other aversive procedures, VS has its drawbacks. First, it is not fully portable. If one experiences deviant urges outside the treatment setting, it is not possible to conduct an actual VS trial. By contrast, adult offenders who have received olfactory aversion routinely carry ammonia ampules for use in vivo. For JSOs, the use of the VS wallet card or a self-directed covert sensitization trial is the preferred coping response. A second disadvantage of VS is an upper limit on the number of trials. Each video vignette can only be shown three or four times before its potency is reduced. This is not the case with olfactory aversion, electric shock, or the satiation techniques. A third disadvantage of VS is that the existing set of videos is restricted to use with child molesters. Perpetrators with older victims or those afflicted with another paraphilia would not likely identify with the offenders or situations portrayed in the existing video series.

Those who utilize conditioning methods with adults may not be terribly impressed with the post-VS arousal levels. Three factors need to be considered. First, the group means obtained herein are not necessarily representative of individual youth performance. A significant minority of subjects showed no change following VS and arousal levels remained high, sometimes at 100%. Second, because adolescents tend to detumesce slowly (compared to adults) stimulus onset occurred at 3 millimeters (approximately 10% of a full erection) if more than two minutes had elapsed and a youth had not returned to baseline. This practice artificially inflated some scores. Third, adolescents become more highly aroused more quickly and to a wider range of stimuli than do adults. This is especially true of younger teens who comprised about half the sample (Kaemingk, Koselka, Becker, & Kaplan, 1995). No sample of adult sex offenders has ever produced a pretreatment mean of over 90% on any audio stimulus or for that matter a mean of 39% to a neutral (party) scene as was obtained in the present study.9 Based on visual stimuli, self-report, and offense history, most JSOs have arousal patterns that are far better crystallized than the audio cues represent. Put differently, the measures that yielded the least favorable outcome were also those with the least convergent and discriminant validity (Weinrott, 1995b).

What little can be gleaned from a three-month followup is positive. There was no indication that the effects of VS dissipated on either phallometric or self-report measures. On the contrary, deviant arousal remained at post-treatment levels or continued to decline. Predicting the pattern beyond three-months is entirely speculative. Parenthetically, there have been no phallometric studies of JSOs that included long-term followup testing subsequent to any form of intervention. Indeed, the present three-month followup assessment represents the longest published to date.

Vicarious sensitization was designed for dissemination. This is especially important because methods developed in research settings almost always produce weaker effects in clinics. Those that have been adopted with the most success tend to be a) behavioral, b) highly structured, and c) focused on a specific symptom or target (Weisz, Donenberg, Han, & Kaunekis, 1995). VS meets all three criteria. A detailed clinical manual accompanies each system and the configuration now in distribution is entirely automated.10 The presentation of one's audio scenario and aversive vignettes is computer controlled; an operator/therapist need only identify the youth and the vignettes to be presented during the upcoming session. Everything after that occurs automatically. The audio scenario is written onto the hard disk, the video vignettes are on laser disks, and all of this material is transmitted to the youth via a virtual reality headset. Sessions can be held in a regular office or small conference room. A therapist need not be present, which makes VS very cost-effective. Amortizing equipment and software costs over 5 years, and including therapist time to write crime scenarios, 25 JSOs can be treated for about $350 each.

References

Abel, G. G., Mittelman, M. S., & Becker, J. V. (1985). Sexual offenders: Results of assessment and recommendations for treatment. In H. Ben-Aron, S. I. Hucker, & C. D. Webster (Eds.), Clinical criminology (pp. 191-205). Toronto, Ontario, Canada: M. M. Graphics.

Ageton, S. (1983). Sexual assault among adolescents. Lexington, MA: Lexington Books.

Barbaree, H. E., & Mewhort, D. J. K. (1994). The effects of z-score transformation on measures of relative erectile response strength: A re-appraisal. Behavior Research and Therapy, 32, 547-558.

Becker, J. V., Cunningham-Rathner, J., & Kaplan, M. S. (1986). Adolescent sexual offenders: Demographics, criminal and sexual histories, and recommendations for reducing future offenders. Journal of Interpersonal Violence, 1, 431-445.

Becker, J. V., & Kaplan, M. S. (1988). The assessment of adolescent sexual offenders. Advances in Behavioral Assessment of Children and Families, 4, 94- 118.

Davis, G. E., & Leitenberg, H. (1987). Adolescent sex offenders. Psychological Bulletin, 101, 417-427.

Emerick, R. L. (in press). Providers manual of treatment of sexual offenders. Beverly Hills, CA: Sage.

Freeman-Longo, R. E., Bird, S., Stevenson, W. F., & Fiske, J. A. (1995). 1994 Nationwide survey of treatment programs and models. Brandon, VT: Safer Society.

Groth, A. N., Longo, R. E., & McFadin, J. B. (1982). Undetected recidivism among rapists and child molesters. Crime and Delinquency, 28, 450-458.

Harter, S. (1988). Manual for the self perception profile for adolescents. University of Denver.

Hunter, J. A. (1991). Personal communication.

Hunter, J. A., & Goodwin, D. W. (1992). The clinical utility of satiation therapy with juvenile sexual offenders: Variations and efficacy. Annals of Sex Research 5, 71-80.

Hunter, J., & Santos, D. (1990). The use of specialized cognitive-behavioral therapies in the treatment of adolescent sexual offenders. International Journal of Offender Therapy and Comparative Criminology, 34, 239-247.

Kaemingk, K. L., Koselka, M., Becker, J. V., & Kaplan, M.S. (1995) Age and adolescent sexual offender arousal. Sexual Abuse, 7, 249-257.

Knopp, F. H., Rosenberg, J., & Stevenson, W. (1986). Report on nationwide survey of juvenile and adult sex-offender treatment programs and providers, 1986. Orwell, VT: Safer Society.

Lakey, J. F. (1994). The profile and treatment of male adolescent sex offenders. Adolescence, 29, 755-761.

Laws, D. R., Gulayets, M. J., & Frenzel, R. R. (1995). Assessment of sex offenders using standardized slide stimuli and procedures: A multisite study. Sexual Abuse: A Journal of Research and Treatment, 7, 45-66.

Laws, D. R., & Marshall, D. R. (1990). A conditioning theory of the etiology and maintenance of deviant sexual preference and behavior. In W. L. Marshall, D. R. Laws, & H. E. Barbaree (Eds.),Handbook of sexual assault(pp. 209-229). New York: Plenum.

Longo, R., & McFadin, B. (1981). Sexually inappropriate behavior: Development of the sexual offender. Law and Order, 3, 21-23.

Malamuth, N. M. (1986). Predictors of naturalistic sexual aggression. Journal of Personality and Social Psychology, 50, 953-962.

Malamuth, N. M., Heavey, C. L., & Linz, D. (1993). Predicting men's antisocial behavior against women: The interaction model of sexual aggression. In G. C. Nagayama Hall, R. Hirschman, J. R. Graham, & M. S. Zaragoza (Eds.), Sexual aggression: Issues in etiology, assessment, and treatment (pp. 63-97). Washington, DC: Taylor and Francis.

Malcolm, P. B., Andrews, D. A., & Quinsey, V. L. (1993). Discriminant and predictive validity of phallometrically measured sexual age and gender preferences. Journal of Interpersonal Violence, 8, 486-501.

Marshall, W. L., & Barbaree, H. E. (1988a). An outpatient treatment program for child molesters. Annals of the New York Academy of Sciences, 528, 205-214.

Marshall, W. L., & Barbaree, H. E. (1988b). The long term evaluation of a behavioral treatment program for child molesters. Behavior Research and Therapy, 26, 499-511.

National Council on Juvenile and Family Court Judges (1993). The revised report from the National Task Force on Juvenile Sexual Offending, 1993, of the National Adolescent Perpetrator Network. Juvenile and Family Court Journal, 44, 1- 120.

O'Shaughnessy, R., McBride, M., Gretton, H., & Hare, R. (in preparation). Psychopathy and recidivism in adolescent sex offenders.

Quinsey, V. L., & Earls, C. M. (1990). The modification of sexual preferences. In W. L. Marshall, D. R. Laws, & H. E. Barbarbee (Eds.), Handbook of Sexual Assault (pp.279-295). New York: Plenum.

Quinsey, V. L., Chaplin, T. C., & Carrigan, W. F. (1980). Biofeedback and signalled punishment in the modification of inappropriate sexual age preferences. Behavior Therapy, 11 , 567-576.

Quinsey, V. L., Rice, M. E., & Harris, G. T. (1995). Actuarial prediction of sexual recidivism. Journal of Interpersonal Violence, 10, 85-105.

Rouleau, J., Abel, G. G., Mittelman, M. S., Becker, J. V., & Cunningham-Rathner, J. (1986). Effectiveness of each component of a treatment program for non- incarcerated pedophiles. Paper presented at the National Institute of Mental Health Conference on Sex Offenders, Tampa, FL.

Ryan G., & Lane, S. (1991). Juvenile Sexual Offending: Causes, Consequences, and Correction. Lexington, MA: Lexington Books.

Salter, A. C. (1988). Treating child sex offenders and victims. Newbury Park, CA: Sage.

Schram, D. D., Milloy, C. D., & Rowe, W. E. (1991). Juvenile sex offenders: A follow- up study of reoffense behavior. Seattle, WA: Washington State Institute for Public Policy.

Serin, R. C., Malcolm, B. P., Kahnna, A., & Barbaree, H. E. (1994). Psychopathy and deviant sexual arousal in incarcerated sexual offenders. Journal of Interpersonal Violence, 9, 3-11.

Steen, C. (1993). The Relapse Prevention Workbook for Youth in Treatment. Brandon, VT: Safer Society Press.

U.S. Department of Justice. (1993). Age-specific arrest rates and race-specific arrest ratesfor selected offenses 1965-1992 (Federal Bureau of Investigation Uniform Crime Reports, December 1993).

U.S. Department of Justice. (1979-92). Criminal victimization in the United States: Anational crime victimization survey report (Crime Surveys: NCJ-145125, National Samples Report Nos. ICPSR 8608, ICPSR 8864, and ICPSR 8625). Annapolis Junction, MD: Bureau of Justice Statistics Clearinghouse.

Weinrott, M. R. (1996). Juvenile sexual aggression: A critical review. Boulder, CO: Center for the Study and Prevention of Violence.

Weinrott, M. R. (1995a). Manual for administration of vicarious sensitization. Eugene, OR: Northwest Media, Inc.

Weinrott, M. R. (1995b). Does it make sense to conduct phallometric assessments of adolescent males? Paper presented at the annual convention of the Association for the Treatment of Sexual Abusers, New Orleans, LA.

Weisz, J. R., Donenberg, G. R., Han, S. S., & Kaunekis, D. (1995). Child and adolescent psychotherapy outcomes in experiments versus clinics: Why the disparity? Journal of Abnormal Child Psychology, 23, 83-106.

Author Note:

This research was supported by the Small Business Innovation Research Program (DHHS) Grant R44 MH44952 and administered by the National Institute of Mental Health. The authors would like to acknowledge the cooperation of the Washington State Department of Juvenile Rehabilitation and the collaborative efforts of Tim Kahn, M.A., and Kirk Johnson, Ph.D.

Requests for reprints and correspondence should be directed to Mark R. Weinrott, Ph.D., Harbor Square Southwest, 5520 SW Macadam Avenue, Suite 180, Portland, OR 97201.

1. To be considered eligible for VS a youth had to obtain at least 50% of a full erection to a video or audio stimulus depicting a young child.
2. Eleven of these 24 boys moved out of the catchment area or were transferred from Echo Glen to another institution or residential program. Of those 13 who actually quit of their own accord, seven did so prior to the first VS session and six after one or more sessions of VS. There were no differences in demographic characteristics or offense histories between the 24 dropouts and the boys who completed the project.
3. After the entire series of vignettes had been shown once, only those vignettes that received a subjective anxiety rating of 6 or higher on a ten-point scale were presented in the second round. Similarly, only vignettes that were rated 6 or higher in round two were presented in round three.
4. 4. A more detailed description of the VS treatment protocol is available elsewhere (Weinrott, 1995a).
5. Parent ratings were obtained for 40 of 48 youths residing in the community. Parents of youths who were institutionalized had too little contact with their sons to form specific impressions of VS.
6. Power was diminished below the conventional 0.8 level due largely to inclusion of a third assessment in which no group differences were expected.
7. This is a much higher mean level of arousal than similar audio scenarios have elicited in adult sex offenders.
8. This parent also reported that her son communicated nothing to him/her about his experience with VS.
9. The corresponding means for videotapes and slides were 10% and 14% respectively.
10. Equipment, stimulus materials, and detailed instructions for administering VS are available from Northwest Media, Inc., P.O. Box 56, Eugene, OR 97440. (800) 777-6636.