Sex Offender Assessment

Although clinical psychological assessment is generally expected to be specific to particular interventions with demonstrated efficacy, there is insufficient empirical evidence on which to prescribe clinical assessment practice with sex offenders. The best strategy is to examine interventions that target personal and circumstantial characteristics empirically related to commission of sex offenses or to recidivism among sex offenders. The most reliable and robust empirical differences between sex offenders and other people pertain to sexual preferences. The best available assessments of the risk of recidivism are provided by actuarial systems that include indicators of deviant sexual preferences and of persistent antisociality across the life course. The Hare Psychopathy Checklist is the best available measure of such antisociality. Equivocal evidence supports the use of some assessments of specific attitudes and symptoms. This entry discusses the psychological assessment of men who have sexually assaulted women and children; much of the discussion also applies to assessment of adolescent male sex offenders.

Forensic Psychological Practice with Sex Offenders

At the time of this writing (early 2007), crucial lacunae affect clinical practice with sex offenders. Despite several decades of research, there are no generally accepted scientific explanations for sex offending. There is also no empirically conclusive evidence about what specific interventions, if any, reduce the likelihood of subsequent offending among sex offenders. Regarding psychological treatment, the most authoritative and comprehensive meta-analysis concluded in 2002 that the available evidence suggested that some psychological therapies were effective, but a firm conclusion could not be drawn. A glaring problem highlighted by meta-analyses is the absence of useful data pertaining to which specific changes in psychological or clinical constructs induced by therapy were responsible for reductions in recidivism (if indeed any had occurred). The intervening 5 years has seen no improvement. Most of the experts’ efforts over the past several years appear to have been devoted to debating, with no evident resolution, whether existing evidence convincingly demonstrated treatment effects. Although novel therapies were recommended and tried, no new effects of sex offender treatment were demonstrated. Consequently, forensic psychologists planning interventions are not in a strong empirical position to know what clinical constructs ought to be assessed for sex offenders. The most appropriate course for psychologists in such circumstances is to implement interventions that specifically target those ostensibly changeable characteristics that distinguish men who commit sex offenses from those who do not or that distinguish those sex offenders who recidivate from those who do not. The intensity of intervention (including supervision and, in the very highest risk cases, incapacitation) should be related to the measured risk of recidivism. Assessment would generally then give greatest priority to assessing this risk of recidivism and the measurement of those constructs selected for intervention.

Some historical and sociodemographic variables are clearly related to offenders’ status as sex offenders or to sex offenders’ risk of recidivism but will mostly be addressed in the later section on risk assessment because their unchangeable nature means they are unlikely to form the basis of clinical psychological intervention (e.g., past sex offending history, past history of other criminal conduct, having experienced abuse, neglect, or abandonment as a child, etc.). Similarly, recent research has strongly suggested that various neurophysiological and neuroanatomical measures are related to sexual offending against children, but again, it would currently be unclear what could be done about such problems in clinical intervention, and therefore, they would be expected to have lower priority in typical clinical assessment with sex offenders. Conversely, other psychological constructs, especially intellectual ability as one example, do not distinguish sex offenders and are very unlikely to be a target for intervention but would usually be assessed because they are probably relevant to program assignment and design. Acute symptoms of psychosis might be another rarer example in this category.

What are the psychological characteristics that distinguish sex offenders from other men or are related to recidivism among sex offenders? There is no doubt that the two most important domains are sexual preferences and life-course-persistent antisociality; these are discussed at greater length in the next two sections. Several other variables are clearly relevant, especially the age and sex of victims and the victim-offender relationship. Sex offenders who target males and adults represent greater than average risk, while those who have targeted only children in their own families represent lower than average risk. Thus, thorough investigation of such details pertaining to all victims, involving collateral sources is probably essential. Although not specific to sex offending, it is probable that assessment for substance abuse problems is relevant to clinical services with all sex offenders. Evidence supports efforts by forensic psychologists supervising sex offenders under conditional release to assess fluctuations in hostility, anger, and noncompliance, and to monitor (via direct observation and collateral reports) access to potential victims.

Much more equivocal evidence relates to the assessment of several other domains including attitudes, beliefs, and symptoms. In this category are assessments of specific justifications and excuses for sexually coercive behavior (e.g., endorsing the idea that victims deserve or enjoy being sexually assaulted) or attitudes supporting generally exploitative and selfish conduct. Similarly, lack of social competence and clinically significant depression have been reported to be relevant in some studies but disconfirmed by others. Polygraphy has been recommended as an adjunct assessment for sex offenders, but evidence to support its use is sparse and equivocal. Some clinical constructs can, on the basis of available evidence, be ruled out as worthwhile assessment concerns among sex offenders. Based on available evidence, self-esteem, anxiety, specific victim empathy, loneliness, denial, and insight would be difficult to support as targets for assessment and intervention among sex offenders. Novel theories about the causes of sex offending might imply new approaches to intervention that would, in turn, imply new ways of assessing these constructs or the measurement of psychological characteristics not yet entertained among sex offenders. Such novel interventions could be ethically attempted as a test of such novel explanations but only in the context of a rigorous outcome evaluation. Comprehensive discussions of the psychometric and other properties of the many formal assessment tools for these characteristics as well as advice about sources of information and interviewing tactics are provided in the recommended readings and related entries.

Assessment of Sexual Preferences

The only assessment shown to reliably distinguish sex offenders from other men (especially offenders without sex offenses) and also to predict recidivism among adjudicated sex offenders is phallometrically evaluated sexual preferences. Phallometry is the measurement of penile tumescence in response to controlled stimuli. Research clearly shows that, on average, men who have sexually assaulted prepubescent children exhibit relatively greater erectile responses to stimuli that sexually depict children (versus adult stimuli) compared with men without such sexual histories. Such “pedophilic” sexual test results are also related to the likelihood of recidivism among such sex offenders. Similarly, several meta-analyses have demonstrated that men who have committed coercive sexual assaults against women exhibit relatively greater erectile response to depictions of rape and violence (vs. depictions of consensual sex) compared with men without such sexual histories, and there is evidence that such test results are related to recidivism. Although the theoretical meaning of these robust and reliable relationships pertaining to sexual preferences has not been fully settled, it seems inescapable that the scientific explanation of sex offending will at least partly depend on them.

There are no fully standardized phallometric assessment packages, but several procedural, instrumentation, and interpretation issues have been worked out. Most important, validity is enhanced with relative measurement—for example, comparing each man’s greatest mean response to a prepubescent child category with his response to his largest adult category or comparing his largest response to a violent category with his mean response to consenting sex. Responses to a single stimulus or category are much less informative. Validity is also improved by having multiple stimuli in each category, using ipsative scoring to remove between-subject variability in overall response magnitude, using stimuli in which the brutality of the deviant sexual behavior is emphasized, and including sexually neutral stimuli to aid in the detection of attempts at dissimulation. Validated stimuli include still pictures and aural stories; video materials have not been shown to be valid in forensic assessment. Other strategies to detect and foil faking have also been reported. Phallometry is probably more specific than sensitive—a deviant result is more informative than a nondeviant profile. There have been reports of low test-retest reliability for phallometric testing, but the well-established discriminate and predictive validity implies that test-retest is probably an inappropriate index of reliability for this form of assessment. Comprehensive discussions of the best ways to conduct phallometric measurement for research and clinical evaluation have been provided elsewhere (see the References: and related entries).

Obviously, phallometric assessment requires specialized techniques and might not be suitable in all forensic practice with sex offenders. In nonclinical populations, self-reported sexual preferences can be valid, but self-report is unlikely to be trusted in most forensic work. There is evidence that a simple count of relevant characteristics of past sexual victims (a male victim, multiple child victims, a victim under the age of 12, and an unrelated victim) is closely related to pedophilic sexual preferences. Of course, by definition, such a measure, though a valid index of preferences, cannot detect changes in such interests. Also, there is no parallel index for sexual interest in coercive and violent sex directed toward adults.

Researchers have attempted a variety of other cognitive-behavioral techniques to detect deviant sexual interests. For example, a picture of a naked person in a man’s most sexually preferred category causes maximal interference (compared with pictures from nonpreferred categories) in some speeded concurrent mental tasks. As another example, when men can control how long they look at pictures, covertly recorded viewing times have also been found to be related to sexual preferences—for example, heterosexual men spend the most time viewing pictures of adult women. Some of these latter measures are still in the development stages, while others have been marketed to forensic clinicians. However, the empirical basis to support their ability to detect sexual interest in children is scant in the context of forensic assessment, and there are no data to support their use to detect interest in the sexual coercion of adults or in the assessment of sex offenders’ risk of recidivism.

Risk Assessment among Sex Offenders

Many forensic psychologists take on the task of assessing the risk of recidivism represented by sex offenders. Much of this work occurs under the aegis of statutes mandating preventative detention for “sexually violent persons,” “dangerous offenders,” or “dangerous people with severe personality disorders” in various English-speaking jurisdictions. This form of assessment is frequently undertaken outside any other clinical responsibility to provide psychological interventions or services. In this context, the assessment of recidivism risk among sex offenders has been controversial, with some commentators saying that psychologists’ participation is unethical. This condemnation has been partly based on the inaccurate assertion that sex offenders are generally very unlikely to recidivate. In fact, the best long-term data indicate that approximately 40% of adult male sex offenders released from secure custody will be apprehended for subsequent sexually violent crime. Most forensic psychologists would probably espouse the view that assessing the risk of sex offend-ers can be ethically conducted as long as psychologists’ practice is in accord with the best available empirical evidence. In that regard, and in contrast with the scientifically less fruitful field of sex offender therapy, there has been clear recent progress in empirically based risk assessment among sex offenders.

This recent empirical progress has occurred in the context of a much larger literature on assessment, in general establishing (beyond any responsible debate) the superiority of actuarial assessment over informal, intuitive, subjective methods. The latter approach occurs when the assessor makes the decision about what information to select, combine, and process. In contrast, the actuarial method eliminates human judgment about the selection of relevant assessment information and about how to combine that information—assessment results depend only on empirically measured relationships between data and the outcome of interest. Several well-validated and replicated actuarial assessments have been developed for use among adjudicated adult sex offenders—men in custody for sexual assaults against women or children. Consequently, psychologists could not defend failing to use such a tool in assessing the risk of recidivism for such sex offenders.

A few psychologists have claimed that instructing clinicians to insert an idiosyncratically determined amount of unaided clinical intuition at the end of the assessment process confers an advantage in predictive accuracy, and they have promulgated nonactuarial schemes for doing so with sex offenders. This promulgation was accompanied by no evidence in support of these claims. The few studies that have since examined these claims among sex offenders show little evidence overall of an improvement due to the addition of unaided clinical intuition, which rather appears to worsen assessment reliability when applied to offenders in general. Because of the overwhelming evidence against the use of unaided clinical intuition in general and the ready availability of actuarial systems for sex offenders, use of such schemes could not be recommended. It should be noted that no actuarial systems have been developed specifically for juvenile sex offenders. However, some of the samples on which the existing actuarial assessments are based included sex offenders who were adolescents at the time of their referral offenses, even though they were adults when followed up. Structured schemes for juvenile sex offenders have been published with some evidence of predictive accuracy.

The available actuarial assessments for sex offender recidivism are moderately to highly intercorrelated, meaning that they do not rank order sex offenders identically. Some of this apparent disagreement is due to differences in the operational definition of recidivism. That is, some researchers have designed actuarial systems to predict only those recidivistic sex offenses labeled as such on police rap sheets. This research strategy is known to miss some officially detected sex offenses, especially the most serious ones (i.e., those involving homicide). It is clear, nevertheless, that the predictive accuracy of all these actuarial systems derives principally from two domains. The first of these was discussed earlier—deviant sexual preferences assessed phallometrically or as indicated by aspects of sex offenders’ offending history (prior sex offenses, multiple sexual victim categories, unrelated and stranger victims, clinical evidence of other paraphilias, etc.).

The second and even more powerfully predictive domain is life-course-persistent antisociality. The best single measure of this for forensic practice is the Hare Psychopathy Checklist (PCL-R), which is an item in some of the available actuarial systems for sex offenders. Other characteristics (some of which are also evaluated in the PCL-R) indicative of this crucial forensic risk assessment construct are a history of criminal and violent offending; parents’ criminality and substance abuse; early age of onset for sexually aggressive, violent, and criminal conduct; aggression and antisocial behavior as a child; disrupted and disturbed family background; failure on prior conditional release; substance abuse; antisocial friends and associates; poor employment stability; hostile and selfish attitudes; quitting or being ejected from sex offender treatment; and meeting the diagnostic criteria for conduct disorder or antisocial personality disorder. Considerable clinical skill is required to assess some of these, but the manner of the assessment and the combination of individual measures should be determined entirely by an actuarial system. Note that these constructs are somewhat redundant, so that an optimal actuarial system need only capture some of them. Instructions on the scoring and interpretation of actuarial risk assessment systems for sex offenders are provided in the recommended readings or suggested related entries.

Conclusions

The sex offenders of greatest forensic concern are those men who sexually assault children and those who engage in sexually coercive offenses against women. There is no accepted scientific explanation of these criminal behaviors and no useful data on specifically what personal or circumstantial changes caused by intervention affect the likelihood of such behavior. As a result, forensic clinicians can rely on very little empirical data to guide assessment when designing treatment. Nevertheless, sufficient evidence exists to show that greatest priority in clinical assessment should be given to comprehensive information on the age, sex, and relationship to all victims; substance abuse; access to potential victims; deviant sexual preferences; and various measures indicative of life-course-persistent antisociality (especially the Hare PCL-R). Available actuarial systems are the most appropriate way to select and combine individual assessment findings to make decisions about sex offenders’ risk of recidivism.

References:

  1. Harris, G. T., & Rice, M. E. (1996). The science in phallometric testing of male sexual interest. Current Directions in Psychological Science, 5, 156-160.
  2. Harris, G. T., Rice, M. E., Quinsey, V. L., Chaplin, T. C., & Earls, C. (1992). Maximizing the discriminant validity of phallometric assessment. Psychological Assessment, 4, 502-511.
  3. Lalumiere, M. L., Harris, G. T., Quinsey, L., & Rice, M. E. (2005) . The causes of rape: Understanding individual differences in the male propensity for sexual aggression. Washington, DC: American Psychological Association.
  4. Prescott, D. S. (2004). Emerging strategies for risk assessment of sexually abusive youth: Theory, controversy, and practice. Journal of Child Sexual Abuse, 13, 83-105.
  5. Quinsey, L., Harris, G. T., Rice, M. E., & Cormier, C. A. (2006) . Violent offenders: Appraising and managing risk (2nd ed.). Washington, DC: American Psychological Association.
  6. Quinsey, L., & Lalumiere, M. (2001). Assessment of sexual offenders against children (2nd ed.). Thousand Oaks, CA: Sage.
  7. Rice, M. E., Harris, G. T., Lang, C., & Cormier, C. A. (2006). Violent sex offenses: How are they best measured from official records? Law and Human Behavior, 30, 525-541.

See also: