Violent and Sexual Offenders

Assessment, Treatment and Management

Jane Ireland, Carol Ireland, Philip Birch, Jane L. Ireland, Carol A. Ireland, Philip Birch, Jane Ireland, Carol Ireland, Jane L. Ireland, Carol A. Ireland, Philip Birch

• Publisher: Routledge
• Year: 2018
• eBook ISBN: 9781315310398
• Topic: Social Sciences
• Subtopic: Criminology
• Index: Social Sciences

Building on the success of the first edition and the growth of research in the field over the past decade, this book offers an authoritative overview of the assessment, treatment, and management of violent and sexual offenders. This new and expanded edition reflects the considerable developments in research and empirical data and captures the increasing breadth of risk assessment approaches, the wider range of empirically based therapies, and the more creative means of considering management.

The second edition captures key developments in this area, with new chapters drawing on a range of pressing contemporary issues, such as female offenders, Internet offenders, terrorists, young people involved in harmful sexual behaviour, and protective factors for aggression. There is also extended coverage of the management of offenders within secure settings and in the community, referring to a wider variety of approaches and the incorporation of technology.

This book will be of considerable interest to academics, practitioners, and students engaged with understanding and/or treating violence and aggression, sex crime, forensic psychology, and the assessment, treatment, and management of offenders.

Section I
Risk assessment – current perspectives

1
Violence risk assessment

Robyn Mooney and Ivan Sebalo

Violence is a prevalent issue in psychiatric and legal settings (Persson, Belfrage, Fredriksson, and Kristiansson, 2017). As a result, risk assessment tools are commonly used to facilitate decision making in this area. Instruments that assess risk of violence are used by psychologists, psychiatrists, nurses, and criminologists to inform decisions about civil commitments, parole and probation, community supervision (Singh, Grann, and Fazel, 2011), and release from psychiatric hospitals or correctional facilities (Guy, Packer, and Warnken, 2012). They may also be used to inform legal recommendations about classifying an offender as a ‘dangerous offender’ or ‘sexually violent predator’ in countries such as the US and Canada (Monahan and Skeem, 2014).

Risk assessment tools and guides aim to determine an individual’s level of risk for future violent behaviour. Some instruments also identify situational triggers that contribute to an individual’s violent behaviour and provide recommendations for how to manage their level of risk (e.g., HCR-20; Douglas, Hart, Webster, and Belfrage, 2013). These assessments serve to benefit both the public and the offender (Strub, Douglas, and Nicholls, 2014). However, when a risk classification is used to justify an extended detention, although detention may be in the best interest of all parties, overestimations of risk may violate the offender’s rights, and underestimations can have dangerous implications for public safety (Douglas, Pugh, Singh, Savulescu, and Fazel, 2017). Therefore, the development and implementation of empirically driven, clinically effective violence risk assessment instruments are of vital importance (Strub et al., 2014). This chapter will describe the various approaches to violence risk decision making and provide an overview of several popular risk assessment instruments and guides for this area of practice.

Approaches to risk decision making

The classic division of risk assessment approaches is an extension of the risk-need-responsivity (RNR) model (Andrews and Bonta, 2010). According to this model, risk factors reflect the variables associated with an individual that are deemed to increase his or her chances of committing an offence. Meanwhile, needs are divided into criminogenic, those in “which assessed change is associated with subsequent criminal behaviour” (p. 27), and non- criminogenic, changes in which are not associated with changes criminal behaviour. The responsivity principle reflects the assumption that the most effective techniques to target criminogenic needs will be those that are adjusted to suit the individual’s abilities, level of motivation, and strengths. Finally, strengths, which reflect characteristics that are deemed to reduce the chances that an individual who possesses them will engage in criminal conduct, are also included in this model.

The risks, criminogenic needs, and strengths from the RNR model roughly correspond to static (or historical) risk factors, dynamic risk factors, and protective factors, respectively. These categories are found at varying rates in different violence risk assessment instruments, which use distinct approaches for deciding an individual’s level of risk (Andrews, Bonta, and Wormith, 2006). The three main approaches are referred to as unstructured clinical judgment, actuarial methods, and structured professional judgment (SPJ) (Doyle and Dolan, 2002). These factors and decision-making frameworks are commonly used to distinguish between three generations of risk assessment.

The first generation refers to unstructured clinical judgment, where a practitioner estimates an individual’s risk of reoffending based on his or her own clinical experience. This approach to risk assessment has been criticised for its over-reliance on the subjectivity of the clinician, and it is widely regarded as both unreliable and invalid (Guy et al., 2012). Meanwhile, the second generation, actuarial methods, introduced the use of research-driven instruments. This approach views risk as a continuous variable, which is directly proportionate to the existence of certain factors that have been empirically established as predictive of target behaviour (Doyle and Dolan, 2002). Most of these predictive factors are static, which means they are unchangeable characteristics (e.g., criminal history; degree of adjustment in school). Hart, Michie, and Cooke (2007, p. 60) explain the essence of the actuarial approach to risk assessment with the following syllogism:

Major premise : In the samples used to construct Test X, 52% of people with scores in Category Y were known to have committed violence during the follow-up period.

Minor premise : Jones has a score on Test X that falls in Category Y.

Conclusion : Therefore, the risk that Jones will commit future violence is similar to the risk of people in Category Y.

While this approach has been demonstrated to lead to higher predictive accuracy than unstructured clinical judgment (Ægisdóttir et al., 2006; Dawes, Faust, and Meehl, 1989; Grove and Meehl, 1996), it has important limitations. Its fixed nature limits it to being sample-specific and incapable of incorporating an individual’s unique risk factors into the calculation of risk when they arise (Hart, 1998). Furthermore, studies have shown that two prominent actuarial tools, the VRAG (Harris, Rice, and Quinsey, 1993) and the Static-99 (Hanson and Thornton, 2000), have surprisingly large 95% confidence intervals (CIs) for predicting violent reoffending on an individual level (Hart and Cooke, 2013; Hart, Michie, and Cooke, 2007). In other words, when these two instruments are used, they lack precision in their predictions about the chances that a particular person will engage in a target behaviour.

The third generation of risk assessment methods, structured professional judgment, capitalised on the success of the previous generation by retaining the empirical basis for establishing risk factors. However, there are also several important differences. Firstly, the selection of risk factors is guided by theory, which facilitates the inclusion of dynamic risk factors that reflect criminogenic needs and can change (e.g., substance use, unemployment) (Andrews et al., 2006). In other words, the development of third-generation instruments did not follow a theoretical model of violent recidivism; rather, a ‘sorting algorithm’ based on statistical significance that can be used to identify strong predictors of recidivism were then turned into items on these risk assessment instruments. This method of instrument construction reflects a common criticism of the second generation regarding its over-reliance on static risk factors. Although they are predictive of violent recidivism, static factors are argued to be insensitive to patient changes (e.g., due to treatment), limiting their ability to inform the rehabilitation process (Andrews et al., 2006; Zamble and Quinsey, 2001). However, the value of adding dynamic factors in risk assessment instruments is yet unclear. While it is argued that the tools that include them appear to have a higher predictive accuracy for recidivism than those which do not, and this is believed to be especially due to dynamic factors (Doyle and Dolan, 2006; Mills and Kroner, 2006), when items are compared on an individual level, those that reflect dynamic risk factors do not always add considerable weight to the predictive accuracy of static ones (Caudy, Durso, and Taxman, 2013; Coid et al., 2011; Giguère and Lussier, 2016; Morgan, Kroner, Mills, Serna, and McDonald, 2013; Philipse, Koeter, van der Staak, and van den Brink, 2006). Furthermore, in recent years, researchers have voiced the need to shift emphasis from concurrent and correlational predictive validity to the aetiology of dynamic risk factors and their causal relationship to reoffending (Cording, Beggs Christofferson, and Grace, 2016; Klepfisz, Daffern, and Day, 2016).

Although the third generation of risk assessment follows the structured professional judgment (SPJ) approach (Doyle and Dolan, 2002), it denotes a combination of the unstructured clinical judgment and actuarial methods, as it uses an empirically constructed ‘checklist’ to guide clinicians’ decisions about a particular individual’s level of risk (Douglas, Cox, and Webster, 1999). This approach takes the process of risk assessment one step further by using the determined level of risk to provide individualised recommendations for treatment interventions and risk management strategies (Singh et al., 2011). SPJ is generally embraced by clinicians as both structured and flexible (Guy et al., 2012), representing a balance between the inconsistency of unstructured clinical judgment and the rigidity of actuarial methods.

With its synthesis of unstructured clinical judgment and actuarial methods, the SPJ approach is relatively successful at retaining the strengths of its predecessors while improving upon their deficiencies. However, some shortcomings remain. This approach has been criticised for allowing too much assessor subjectivity (Quinsey et al., 2006). Indeed, there is some evidence that when clinicians use their discretion to categorise risk nominally (low, moderate, or high) rather than calculating it numerically, this results in an overestimation of risk (Mills and Kroner, 2006) and reduces the instrument’s predictive validity.

Overall, the actuarial and SPJ approaches (thus, second and third generations) are both moderately capable of predicting an individual’s risk for future violence (Guy et al., 2012). However, unlike the SPJ approach, actuarial methods fail to provide guidance on how to treat and manage an individual’s personal risk factors. Consequently, SPJ has emerged as the most effective of these three approaches to risk assessment.

Popular risk assessment instruments and guides for violence

This section will provide an overview of some of the most commonly used second- and third-generation violence risk assessment instruments and guides.

Psychopathy Checklist-Revised (PCL-R)

The Psychopathy Checklist Revised (PCL-R; Hare, 2003) is a second-generation diagnostic tool developed for the detection of clinical psychopathy in adults. Psychopathy is a personality construct that taps into a variety of affective, behavioural, and interpersonal characteristics, including antisocial behaviour and impaired empathy (Hare, 2003). These characteristics are commonly split into two factors: Factor 1 represents interpersonal and affective features of psychopathy, while Factor 2 refers to elements of antisocial behaviour and lifestyle. The PCL-R has been applied to many populations, including men and women with substance use disorders, individuals with personality disorders, sex offenders, and offenders without mental disorders (Coid et al., 2011). It is used for the assessment of psychopathy in many countries worldwide, including the US, Canada, Scotland, England, Spain, Portugal, Germany, Belgium, Denmark, Norway, Sweden, Finland, and Brazil. Although this instrument was not designed for use in violence risk assessments per se, it is often described as a moderately good predictor of both violent and general recidivism (Coid et al., 2011; Hare, Clark, Grann, and Thornton, 2000; Leistico, Salekin, DeCoster, and Rogers, 2008), and it correlates at moderate to high levels with violence risk assessment instruments and guides, such as the HCR-20 and VRAG (Douglas, Vincent, and Edens, 2006). However, the empirical findings regarding its ability to predict future violent recidivism are far from unanimous. A meta-analysis (Walters, 2003) concluded that Factor 2 is significantly more predictive of general and violent recidivism than Factor 1. This finding has been reiterated in more recent studies (e.g., Olver and Wong, 2015; Yang, Wong, and Coid, 2010). Meanwhile, other studies have demonstrated that Factor 1 and Factor 2 are equally effective at predicting violent recidivism in samples of Swedish offenders with personality disorders (Grann, Långström, Tengström, and Kullgren, 1999) and schizophrenia (Tengström, Grann, Långström, and Kullgren, 2000).

In addition to these mixed findings, the predictive validity of psychopathy as measured by the PCL-R has not been extensively validated in samples such as non-offenders, civil psychiatric settings, community samples, and ethnic minorities (Douglas et al., 2006). Furthermore, psychopathy is just one of several predictors of violence risk, and an individual may exhibit a high level of risk in the absence of psychopathic traits. Nonetheless, a PCL-R score’s ability to predict violence when used as a diagnostic tool has led to its inclusion as a risk item on guides designed specifically for the assessment of risk, including the VRAG (Harris et al., 1993) and the HCR-20 (Webster, Eaves, Douglas, and Wintrup, 1995).

Violence Risk Appraisal Guide-Revised (VRAG-R)

The original version of the Violence Risk Appraisal Guide (VRAG) was developed based on a sample of 618 male offenders, more than half of whom had been admitted to psychiatric hospital in Canada. Stepwise discriminant analysis of biographical data yielded 12 variables as significant correlates of violent recidivism, defined by this instrument as “new charge for a criminal offense against persons or return to maximum security institution for violent behaviours” (Harris et al., 1993: 322). The sum of standardised scores for each participant significantly correlated with violent recidivism and the newly formed 12-item instrument was found to have sensitivity of .60 and specificity of .78 when the base rate of violent recidivism was used as the cut-off score (for a more detailed description of the VRAG’s development, see Quinsey, Harris, Rice, and Cormier, 2006).

The VRAG has been used for over two decades and has been successfully validated across countries and samples – for instance, in male forensic patients in the UK (Doyle, Carter, Shaw, and Dolan, 2012); male and f...

Table of contents

• Cover
• Title
• Copyright
• Dedication
• Contents
• List of tables and figures
• List of contributors
• Foreword
• Section I Risk assessment – current perspectives
• Section II Clinical assessment – current perspectives
• Section III Treatment – current perspectives
• Section IV Management – current perspectives
• Index