Motivation and emotion/Book/2016/Sexual offender risk assessment

Overview
Recidivism is the reversion to criminal behaviour by an individual who was previously convicted of a crime (Maltz, 2001). Within the field of criminology and sexual offences, the most common definition of recidivism involves reconviction of the same crime, or same style of crime (e.g., sexual versus non-sexual, violent versus non-violent). The rate of reconviction for sexual offenders is often considered to be the most convenient and verifiable measure of recidivism, and is the definition used here, although it is also the lowest estimate of recidivism and is likely to significantly under-represent the actual rate of behaviour. Some researchers have argued that the rate of re-arrest is a more accurate estimate of actual behaviour which, in some samples, is reported to be double the rate of reconviction (Loucks, 2002). The actual rate of recidivism may be a great deal higher, with approximately 0.9% - 1,383 (ABS, 2006a) of 143,900 (ABS, 2006b) - of reported sexual assault perpetrators in Australia in 2006 being convicted.

Predicting recidivism in sexual offenders has been the subject of considerable research, with some commentators describing it as "notoriously difficult and ... often inaccurate" (Gelb, 2007, p. 30). Prediction of re-offence remains largely based on the examination of unchanging (static) risk factors through the use of actuarial tools such as the Static 99R (Hanson & Thornton, 2000), Static-2002R (Hanson, Helmus, & Thornton, 2010), and MnSOST-R (Epperson et al., 2000) which have "a known degree of predictive accuracy (in the moderate range)" (Hanson & Morton-Bourgon, 2005, p.3). However, recent meta-analysis has found that the use of instruments examining both static and changeable (dynamic) risk factors, such as the The Violence Risk Scale: Sexual Offender Version (VRS:SO; Beggs & Grace, 2010), continue to gain popularity (Hanson & Morton-Bourgon, 2009). Theories underpinning the motivational drives of sexual offenders have been proposed over time, such as the Groth Typologies (Groth, 1979), but none currently have strong research support for their predictive ability.

Sexual Offences in Law
Legal definitions of a sex offence vary widely both between and within countries. As such, no unilateral definition exists as to what they constitute, although broad terms exist to describe typically prohibited sexual behaviour. Examples include:
 * Voyeurism, observing another person in a sexual manner
 * Frotteurism, touching or rubbing a body part against another person without their knowledge or consent
 * Prostitution, trading sexual acts for recompense (e.g. money or drugs)
 * Child Pornography, the production, distribution, possession, or consumption of media containing sexualised depictions of those below the age of consent
 * Rape, sexual penetration of another person without their consent
 * Child Sexual Abuse, the use of a child for sexual stimulation
 * Incest, sexual engagement between people with a consanguineous relationship

Courts will often utilise a standard set of principles which need to be met before a person can be prosecuted with a sexual offence. As an example, to be convicted of a sexual offence within Australia, four elements must be established: Definitions of how consent is demonstrated or assumed (i.e., active versus passive consent) within sexual contact is subject to ongoing debate in Australian Law (Fileborn, 2011).
 * 1) Actus Reus: Sexual activity occurred, and it occurred without the consent of the victim
 * 2) Mens Rea: The defendant was aware that the victim did not consent to the act
 * 3) Voluntariness: The defendant’s actions cannot have been involuntary (e.g. sleepwalking, reflex actions). Self-induced intoxication is not considered evidence of involuntariness
 * 4) Temporal Coincidence: That Actus Reus and Mens Rea occurred at the same time

Motivational Theories of Sexual Offences
Due in part to the widely varied nature of sexual offences, literature focusing on the prediction of recidivism in sexual offenders has historically tended to examine specific predictive utility within subgroups, rather than looking for a grand or unifying theory of all sexual offences (Roberts, Doren, & Thornton, 2002). Where examined, the aetiology of sex offences is often cited as poor family environments characterised by abuse and neglect, which leads to poor attachment and guidance. This in turn creates opportunity for impersonal and aggressive attitudes towards intimacy and sexuality, and the development of attitudes allowing for non-consenting sexual acts (Ward & Siegert, 2002; Malamuth, 2003). Despite the prevalence of this view, evidence to back it up has been poorly supported in the literature. For example, a meta-analysis (N = 5,711) found that being sexually abused as a child was not significantly related to sexual recidivism (Hanson & Bussière, 1998; Hanson & Morton-Bourgon, 2005). Several more specific theories have been suggested that have sought to explain parts of sexual offender motivation, including; The Groth Typologies (Groth & Birnbaum, 1979), Knight's Unified Theory of Sexual Coercion (Knight, 1999), the theory of Sexual Offender Pathways (Hudson, Ward & McCormack, 1999), and the Integrated Theory of Sexual Offending (ITSO; Ward & Beech, 2006).

The Groth Typologies
The Groth Typologies, initially proposed in 1979, were created to explain the motivational drive for rape. Considered by some researchers to be "the most famous" (Lin, Maxwell, & Barclay, 2000, p. 223) of the early motivational explanations of rape and child molestation, they arise from Freud's interpretation of Drive Theory. Groth explains both types of sexual offence as addressing largely non-sexual needs in the context of personality disturbance, such as the absence of closeness and a general experience of inadequacy in relation to others, particularly in a sexual realm (Chesire, 2004).

His formulation contained three specific types of rapist; the power rapist (55%), the anger rapist (40%), and sadistic rapists (5%) (Groth & Birnbaum, 1979). Follow up studies have found a degree of similarity in the representation of these types within other cultures (e.g., Hsu & Ma, 1992).

The Power Rapist
Power rapists are those who rape as a way of exercising control or dominance over another person, with the aim of experiencing sexual fetishism or the confirmation of sexual desirability. This type of rapist is generally more likely to utilise psychological manipulation than physical force.

The Anger Rapist
Anger rapists' motivation is conceptualised as being primarily in terms of hurting, debasing, punishing or disgracing the other person (Lin, Maxwell, & Barclay, 2000). This is accomplished through the use of violence or strength as an outlet for frustration either directly related to the victim or as an outlet for frustration directed at another person.

The Sadistic Rapist
Both the rarest and the most likely to cause death, sadistic rapists are driven by the specific sexual enjoyment of the suffering of their victims. Both the infliction of pain and the rape are typically ritualised, and will sometimes culminate in the murder of the victim.

Child Molesters
Groth also sought to explain child molestation in two types; regressed type and fixated type child molesters. Within this explanation, the regressed type are those who may have previously had age appropriate relationships but due to situational factors that have reduced their confidence and transferred their sexual gratification to those who are less threatening. Alternatively, those with a fixated type attraction are more likely to engage in manipulative behaviour (e.g., grooming), and have had a specific attraction to children throughout life.

Integrated Theory of Sexual Offending
Created by Ward and Beech (2006), the ITSO attempts to combine three domains (biological, ecological nich, neuropsychological) to create a more resilient model to predict sexual offending.

Biological Factors
Biological factors within the ITSO are influenced by genetic and brain development. These factors are the translation of genetically selected needs (e.g. sex, autonomy) through biological drive reinforcement. In such a way the influence of characteristically useful hormones, such as cortisol, serotonin, and dopamine, can impact on an individual's behaviour.

Ecological Niche Factors
Ecological Niche Factors are those which relate to personal circumstances, such as potentially adverse cultural and social influences. These are divided into two general areas; distal factors and proximal factors. In this instance the distal factors refer to the inability of an individual to meet the challenges of their environment, which creates psychological vulnerability to succumb to temptation when opportunity is presented. Proximal factors are those that arise out of the person's situation, such as the advent of war or the death of a spouse. The combination of these two factors presents the highest risk of both offence and recidivism on this dimension.

Neuropsychological Factors
Within the ITSO, neuropsychological factors consist of three domains:

Motivational and emotional factors are those relating to the cortical, limbic, and brainstem structures. Examples given include emotional dysregulation originating from poor role modelling in childhood.

The action selection and control system is associated with the frontal cortex, basal ganglia and the thalamus. These are classically thought of as stable dynamic risk factors, and include impulsivity and inability to adjust plans to changing circumstances.

The perception and memory system is associated with the hippocampal formation and posterior neocortex. Problems in this system lead to problematic interpretations of social engagements and addition to cognitive distortions.

Predicting Recidivism
A large meta-analysis of 61 studies (N = 28,972) conducted by Hanson and Bussière (1998), which is also the most cited article on sex offender recidivism (Soothill, 2010), found that the overall rate of recidivism for sexual offenders after 5 years was 13.4%. This is considered relatively low compared to the overall risk of recidivism in the criminal population, with the largest US study (N = 272,111) reporting 46.9% of all those released from prison reconvicted of any crime within three years, with 67.5% charged with a new offence in the same time (Langan & Levin, 1994).

The field of recidivism risk assessment could be considered to exist in two parts. The most established are those that are based on the identification of correlated, static risk factors, often derived through the meta-analyses conducted by Hanson and Bussière (1998) and Hanson and Morton-Bourgon (2005). Assessment tools utilising this information are known as actuarial (statistical) risk assessments. The second is the assessment of changeable (dynamic) risk factors, which can be further classified into stable and acute dynamic predictors (Perkins et. al., 1998).

Previous Criminal History
The factor often cited as the most strongly associated with risk of recidivism (Scalora & Garbin, 2003). Within this, sexual criminal history has been one of the strongest predictors of recidivism (Hanson & Bussière, 1998; Quinsey, Lalumiere, Rice, & Harris, 1995), with diversity of the type of sexual offence (Gibbens, Soothill, & Way 1981), total number of previous offences and age at first offence (Hanson & Bussière, 1998) also being positively correlated with offending rates.

Type of offence
The rate of reconviction varies between different types of offender (Soothill, 2010). For example, for child molesters the rate of recidivism for sexual crime is 12.7% and with rapists it is 18.9%, which increased to 36.9% and 46.2% respectively when recidivism was defined as any type of crime (Hanson & Bussière, 1998).

Age and Marital Status
Age is often negatively associated with rates of recidivism (Hanson, 2001), and is incorporated into actuarial measures such as the Static-99 as a negative predictor. Similarly, being single is predictive of higher rates of recidivism (Hanson & Bussière, 1998).

Gender
Female sex offenders are less likely to re-offend once convicted. Cortoni and Hanson (2005) found over a 5 year period that males were reconvicted of another sexual offence in between 13-14% of cases, compared to 1% of female offenders. Additionally, the gender profile of sexual offenders is also predominantly male. In a report to the Sentencing Advisory Council, Dr Karen Gelb (2007) found that of the 1,816 defendants adjudicated for sexual assault in Australia in 2004-2005, only 27 were female.

Dynamic Risk
Dynamic risk factors are those that are amenable to change over time, and as such are often those targeted directly by treatment (Gelb, 2007).

Dynamic (Stable)
Stable dynamic risk factors are those which are considered potentially changeable but relatively stable over time (Perkins et. al., 1998).

Antisocial Orientation: Antisocial personality, as found in those diagnosed with Antisocial Personality Disorder, as well as antisocial traits and a history of rule violation have been found to be associated with sexual recidivism. These are most typically measured by the Hare Psychopathy Checklist (Hare et al., 1990) and the MMPI psychopathic deviate scale. Similarly, measures related to antisocial personality, such as employment instability, substance abuse, and hostility all have small effect sizes (d =.11 to .22) on predicting sexual recidivism (Hanson & Morton-Bourgon, 2005).

Deviant Sexual Interests: Measures of deviant sexual interest are typically associated with recidivism (d = .31), with sexual interests in children, preoccupations with sex (both paraphilic and non-paraphilic), and high femininity scores on the Minnesota Multiphasic Personality Inventory (MMPI) also associated with higher risk (Hanson & Morton-Bourgon, 2005).

Dynamic (Acute)
Acute dynamic risks are those that can rapidly change and are associated with increased or decreased recidivism risk (Andrews, Bonta, & Hoge, 1990), and are commonly used in other types of recidivism risk assessment (Andrews, Bonta, & Wormith, 2004). Despite strong theoretical links between dynamic risk factors and sexual recidivism, according to Hanson and Morton-Bourgon (2009) in a recent review; "The research base for dynamic (causal) risk factors among sexual offenders is ... much less developed." (p. ?{{missing)))

Research completed by Hanson and Harris (1998) showed that within their sample of 208 offenders, recidivists more commonly showed increased anger and distress at the time of the offence, poor social support, poor self-management strategies, and difficulty coping with supervision. Similarly, in a later study, Hanson and Harris (2000) found that within their sample (N = 400 sex offenders), consistently higher rates of recidivism were observed in those who abused drugs and alcohol (r =.17) in addition to a correlation with the amount of substances consumed before re-offending (r = .16). Small differences were found with social influences, with higher negative influences reported in those who re-offended{{explain}}.

Protective Factors and Treatment
Risk assessment for sexual recidivism is focussed primarily on historical and current individual deficits. However, there has been a movement to include in risk assessment factors which relate to an individual's areas of strength. Known as the Desistance Theory, proponents suggest that criminal behaviours, such as sexual offences, occur when individual lack the resources to meet their needs through lawful means. One such group are those that support the Good Lives Model (Ward & Marshall, 2004; Ward & Maruna, 2007), who propose that changes occur in criminogenic behaviour as a process of individual maturation. Despite research in this area being in its infancy, advocates point out established correlations, such as those with age, marriage, work and job stability, and social acceptance, as being reflective of this process (Scoones, Willis & Grace, 2012).

Treatment of Sexual Offenders
Early prison-based treatment of sexual offenders returned poor results, and had generally little ability to reduce rates of recidivism (Furby, Weinrott, & Blackshaw, 1989; Quinsey, Harris, Rice, & Lalumiere, 1993), although results at the time were somewhat mixed (Marshall, et al., 1991). However, a large, more recent meta-analytic review showed stronger outcomes, particularly with therapies focusing on cognitive-behavioural techniques returning a 37% reduction in rates of recidivism (Lösel & Schmucker, 2005). Some criticism of the methodology of included studies has arisen however, with critics pointing out that only a small number of the trials included used either randomised experimental designs (7%), or individual matching or statistical control (9%) thus increasing the chances of selection bias.

A more recent study conducted by Duwe and Goldman (2009) which looked at treatment outcomes in a large (N = 3,440), matched sample of sex offenders released from Minnesota prisons between 1990 and 2003 found strong evidence for their treatment programs. They reported that entering treatment reduced the risk of recidivism by 27%, and completing treatment reduced this by a further 33%. It was noted within the study however that they were unable to account for community based treatment program, which have also been shown to significantly reduce the rate of reconviction (Aytes, Olsen, Zakrajsek, Murray, & Ireson, 2001).

Assessment Tools
Assessment tools for predicting sexual recidivism vary in style and accuracy. Several ways of classifying these exist, with the most common being those proposed by Hanson and Morton-Bourgon (2009). A summary is presented below:

Empirical Actuarial
Defined as tools with explicit items determined in advance, and specific coding procedures which produce a score that can be compared to expected recidivism rates derived from research. The Static-99 (Hanson & Thornton, 2000), Rapid Risk Assessment for Sexual Offence Recidivism (Hanson, 1997), and Minnesota Sex Offender Screening Tool-Revised (Epperson et al., 2000) are examples of such tools. These have been found to be generally the most accurate in predicting recidivism (d = .67).

Mechanical
Similar to the empirical actuarial tools, these tools provide a total score derived from specific coding but do not contain comparative information, or where they do are based on theoretical frameworks rather than specific research.

Adjusted Actuarial

These tools are based on an actuarial or mechanical tool, but contain scope for the administering clinician to adjust the final rating based on their clinical judgement. Factors which can alter the final rating are not necessarily specified in advance. An example of an adjusted actuarial tool is the Sexual Violence Risk-20 (SVR-20; Boer, Hart, Kropp, & Webster, 1997). Such tools can have similar predictive validity as empirical actuarial tools (Hanson & Morton-Bourgon, 2005).

Structured Professional Judgement
Assessors are given a list of pre-determined factors before the assessment, but the method and overall evaluation is left entirely at the discretion of the assessor. The HCR-20 (Webster, Douglas, Eaves, & Hart, 1997) is an example of such a tool.

Unstructured
Risk assessment is based purely on individual case analysis, case conferences, and professional experience. These were the primary method of assessment between 1970 and 1998 (Hanson & Morton-Bourgon, 2009), but due to low predictive reliability are the least used currently.

Conclusion
Sexual offences are a broad category of criminal offences that are differently defined depending on the country or state. Numerous theories have been proposed over time as to the motivational impetus of sex offence behaviour, typically looking at a range of factors in the aetiology and perpetuating factors of such behaviour. However, scientific research has not strongly supported any one model. Assessment of risk has tended to focus on statistical correlates of recidivism, and the formalised assessment of such factors is known as actuarial risk assessment. More recent assessment tools have also sought to include dynamic risk and protective factors, although such research is still in its infancy. Treatment of sexual offenders both in prison and in the community can provide reduction of risk, particularly with cognitive-behavioural treatment models.