Forensic Psychology

by Graham M Davies

  Murphy and Clare (1996) have developed the Fire Setting Assessment Schedule (FSAS) for use with adult fire-setters with ID. The construction of the FSAS was guided by the functional analytical approach to fire-setting proposed by Jackson, Glass, & Hope (1987) in which fire-setting is hypothesised to be associated with a number of psychological functions including the need for peer approval, need for excitement, a need to alleviate or express sadness, mental illness, a wish for retribution and a need to reduce anxiety. The FSAS is comprised of 16 items concerning cognitions and feelings prior to setting fires, such as: “I started fires to make people pay attention and listen to me”; “I started fires to get out of going somewhere or doing something”; “I started fires because I felt angry with people”; and a further 16 items pertaining to the individual’s thoughts and feelings after setting fires, for example: “After the fires began I felt calmer”; “After the fires people took more notice of me”; “I felt happier after the fires”. All items are rated as true or false.

  Murphy and Clare (1996) interviewed 10 adult fire-setters with ID using the FSAS. They found that the participants in their study identified antecedents to fire-setting with more reliability than consequences. The most frequently endorsed antecedents were anger, followed by being ignored and then feelings of depression. This assessment has proven to be clinically useful since its inception, but there has been little further research on its reliability and validity until Taylor et al. (2002) used the FSAS in the assessment and treatment of a group of 14 people with mild-borderline ID to review the effectiveness of a fire-setting programme for this client group. Consistent with the results of Murphy and Clare (1996), Taylor et al. (2002) found that anger, being ignored and depression (in rank order) were the most frequently endorsed items on the FSAS in terms of antecedents to and consequences of participants’ fire-setting behaviour. In a further study of six women with mild ID who had set fires, Taylor, Robertson, Thorne, Belshaw and Watson (2006) also found that anger and depression were the most frequently endorsed items in participants prior to fire raising incidents.

  Murphy and Clare (1996) also developed the Fire Interest Rating Scale (FIRS) for use with clients with ID and histories of fire-setting. The FIRS consists of 14 descriptions of fire-related situations, for example: “Watching a bonfire outdoors, like on bonfire night”; “Watching a fire engine come down the road”; “Striking a match to set fire to a building”. Respondents are asked to rate how they would feel in each situation using a 7-point scale from “most upsetting/absolutely horrible” to “very exciting, lovely, nice”.

  Taylor et al. (2002) developed a series of Goal Attainment Scales (GAS; Kiresuk & Sherman, 1968) to support the clinical evaluation of the responses of fire-setters with mild-borderline ID to a group therapy intervention. Based on their answers in a semi-structured interview, a group therapist and an independent rater used the GAS to score clients in relation to six offence-related treatment targets: (i) acceptance of guilt; (ii) acknowledgement of responsibility; (iii) understanding of victim issues; (iv) understanding of high risk elements of the offence cycle; (v) appropriate expression of emotion; and (vi) ability to form and maintain relationships. Assessors used a five-point scale from (0) “much worse than expected”, through to (4) “much better than expected” for each GAS using operationalised scoring criteria.

  21.4.4 Assessment of Risk for any Future Incidents

  There continue to be important developments in the assessment of risk of future incidents in the field of forensic ID. Fitzgerald, Gray, Taylor, & Snowden (2011) assessed the predictive validity of a number of criminal history variables with 145 offenders with ID. In a follow-up of over two years, they found that the number of previous offences, the number of previous drug offences and a number of previous acquisitive offences were all significantly related to future of offending. A history of alcohol abuse and a history of drug abuse were also related to reoffending. Fitzgerald et al. then used risk assessment developed for use on probation settings containing six variables: gender, current offence, age at first conviction, age at current conviction, number of previous custodial sentences and total number of previous convictions. This risk assessment, based on previous assessments, was highly predictive of future reoffending for these offenders with ID.

  Lindsay, Elliot, and Astell (2004) conducted a study to review the predictive value of a range of previously identified variables in relation to recidivism for 52 male sex offenders with ID. Significant variables to emerge from the regression models were generally similar to those variables identified in mainstream studies. However, certain items such as employment history or deviant victim choice, which had been highly associated with recidivism in studies of mainstream offenders, did not emerge as predictor variables. On the other hand, dynamic variables and those related to start behaviour and attitudes did emerge as strong predictors. The authors considered that this may be an indication of the way in which professionals making up assessments in this field should adjust their perceptions. For example, while fewer individuals with ID have an employment history, they are likely to have alternative regimes of special educational placement, occupational placement and the like, which make up a weekly routine of engagement with society. Non-compliance with this regime did emerge as a significant variable, suggesting that individuals with ID should be judged in relation to their peers.

  Quinsey, Book and Skilling (2004) conducted an assessment of the Violence Risk Appraisal Guide (VRAG) in a 16-month follow up of 58 participants with ID. They found significant predictive value for violent incidents with a medium effect size and that staff ratings of client behaviour significantly predicted antisocial incidents. They also assessed the value of dynamic/proximal risk indicators (see Chapter 18 ). They found that in the month prior to the violent/sexual incident, dynamic indicators of anti-sociality were significantly higher than for those recorded six months prior to the incident. This, they concluded, provided persuasive evidence of the value of dynamic assessment since the increase in dynamic risk factors one month prior to the offence could not be attributed to any bias from the incidents. Once again, the study highlighted the importance of dynamic risk factors for this client group.

  In a more extensive investigation into the VRAG, the HCR-20 (Historical, Clinical, Risk Management-20, see Chapter 18 for details) (Webster et al., 1997) and the Psychopathy Checklist- Revised (see Chapter 18 for details) (Hare, 1991, 2003), Gray, Fitzgerald, Taylor, MacCulloch, & Snowden (2007) compared 145 forensic patients with ID and 996 mainstream forensic patients all discharged from hospital. They found that all of the assessments predicted reconviction rates in the ID sample with the effect sizes as large or larger than the mainstream sample. In a further risk assessment comparison study, Lindsay et al. (2008) employed a mixed group of 212 violent and sexual offenders with ID. They followed participants for one year and found that the VRAG was a significant predictor of future violent incidents, and the Static 99 (Hanson & Thornton, 2000; and see Chapter 18 for description of static risk assessment measures for sexual offenders) was a significant predictor of future sexual incidents. Camilleri and Quinsey (2011) reviewed risk assessment for this client group and also conducted a small study on the VRAG. In total sample of more than 700 participants, they found that the VRAG produced equivalent predictive values for those falling in the ranges of borderline intelligence and mild ID ranges when compared to the rest of the sample. Taylor et al. (2011) have reviewed the psychometric properties of predictive validity of the HCR-20. They found that inter-rater reliability was acceptable at over 80% agreement for all scales and Cronbach’s alpha was acceptable for the H scale (0.75) but low for the C and R scales (0.59 and 0.39 respectively). They also found that the R scale had the highest predictive value in relation to recorded incidents over a period of one year. Based on these studies, we can conclude that the VRAG, the HCR-20 and the Static 99 have utility in forensic ID services.

  Employing a similar design to that of Quinsey, Book and Skilling (200
4), Lindsay et al. (2004b) tested the Dynamic Risk Assessment and Management System (DRAMS) on which staff made daily ratings of clients’ mood, antisocial behaviour, aberrant thoughts, psychotic symptoms, self-regulation, therapeutic alliance, compliance with routine, and renewal of relationships. Ratings were compared between those taken on the day of the incident, the day prior to the incident and for a control day at least seven days distant from an incident. There were significant increases in ratings for the day prior to the incident for mood, antisocial behaviour, aberrant thoughts and DRAMS total score. Steptoe, Lindsay, Murphy, and Young (2008) conducted a larger study on the predictive utility of the DRAMS using the same design. There were highly significant differences, with large effect sizes, between assessments taken one or two days prior to an incident and control assessments conducted at least seven days from an incident. Therefore, dynamic risk assessment seemed to perform well in both concurrent and predictive validity in relation to offenders with ID.

  Lofthouse, Lindsay, Totsika, Hastings, and Roberts (2014) developed a dynamic risk assessment for violence in offenders with ID. The Current Risk for Violence (CuRV) was a brief uncomplicated assessment designed to be used by staff who had experience of people with ID but were untrained in risk assessment. Predictive accuracy for aggression was tested prospectively among 64 adults with an ID and history of aggression. Initial findings revealed good predictive validity over a five-month period: AUC range from .72 to .77.

  In a new approach to risk assessment for sex offenders with ID, Boer, Tough and Haaven (2004) developed the ARMIDILO (see Box 21.1).

  BOX 21.1 The ARMIDILO (ASSESSMENT OF RISK AND MANAGEABILITY OF INTELLECTUALLY DISABLED INDIVIDUALS WHO OFFEND)

  ARMIDILO contains 30 items in four broad categories.

  The first section on stable dynamic environmental and staff factors contains items such as communication amongst staff and attitudes towards sex offenders with ID.

  The second section relates to stable dynamic factors in the offender being rated with items such as compliance with supervision and self-management of sexuality.

  The third section on acute dynamic environmental factors includes items on new supervisory staff and any changes in monitoring

  The final section relates to offender acute dynamic factors and contains items such as changes in social support, changes and substance abuse or changes in sexual preoccupation.

  Blacker, Beech, Willcox, and Boer (2011) assessed the predictive value of the subscales on the ARMIDILO and found that the combined acute sections predicted future incidents with a medium effect size. Lofthouse, Lindsay, Totsika, Hastings, Boer, and Haaven (2013) also evaluated the predictive utility of the ARMADILO with 64 sex offenders with ID who were followed up for six years. They found that the Armadilo total score was highly predictive of future offences. These results would suggest that dynamic variables have significant utility in the prediction of future sexual and violent incidents in offenders with ID. Quinsey, Book and Skilling (2004), Lindsay et al. (2004b, 2008), Blacker et al. (2011), and Lofthouse et al. (2013, 2014), all found that dynamic variables predicted future incidents as well as or better than static variables.

  21.5 INTERVENTIONS WITH OFFENDERS WITH ID

  One of the main developments in the expansion of offender rehabilitation programmes in prisons and probation services has been in the improvement of social problem solving and cognitive skills. The purpose of cognitive skills programmes is to equip offenders with thinking skills that will promote alternative, prosocial means of approaching situations in which the person is at risk of offending. These alternative thinking styles will allow the individual to move out of the habits of offending lifestyle that may be reinforced by criminal thinking styles. Wilson, Bouffard and McKenzie (2005) evaluated the range of cognitive behavioural programmes for offenders. They found that in general, programmes that concentrated on appropriate problem solving skills produced a significant reduction in reoffending (7%–33%). They also concluded that high quality studies with greater programme integrity produced the highest reductions in reoffending. Subsequent reviews (Joy Tong & Farrington, 2006; Pearson, Lipton, Cleland, & Yee, 2002) also concluded that such programmes reduce reoffending by 14%–21%.

  Given that offenders with ID are likely to have deficits in social problem solving due to lower intellectual functioning, these developments have only been extended to this field recently. (Lindsay, Hamilton, Moulton, Scott, Doyle, and McMurran (2011a) have developed a Social Problem Solving and Offence Related Thinking (SPORT) programme designed for ID offenders. They described 10 case studies treated in two groups of five who participated in the programme. Following treatment, participants reported feeling more positive about interpersonal problems, and also were assessed as being less impulsive and less avoidant in social situations. These improvements maintained at three-month follow-up. Despite these positive results, these results can only be considered to be of a preliminary nature. We will now examine interventions for specific groups of ID offenders in more detail.

  21.5.1 Treatment for Anger and Violence

  Within this field, the most common treatment approach for offenders has been behavioural interventions (Carr et al., 2000). One difficulty in employing behavioural approaches with offenders is that they generally require contingencies to be organised in a consistent and reliable fashion, in a controlled institutional environment with a reasonable staff ratio. Such conditions contrast with those in services for offenders with ID who may be relatively high functioning (in relation to other services for people with ID), display infrequent yet very serious aggression and live in relatively uncontrolled community settings. In response to the need for more “self-actualising” treatments that promote generalised self-regulation of anger and aggression, several authors employed cognitive behavioural treatments based on the approach developed by Novaco (1975, 1994). This approach employs cognitive restructuring, arousal reduction and behavioural skills training as well as stress inoculation techniques.

  PHOTO 21.3 Cognitive behavioural anger treatment involving offenders with ID have yielded positive outcomes.

  Source: © Nomad_Soul/Shutterstock

  Taylor (2002), and Taylor and Novaco (2005), have reviewed a number of case studies and uncontrolled group anger treatment studies involving individual and group therapy formats, incorporating combinations of cognitive behavioural techniques including arousal reduction, skills training and self-monitoring. Generally, these case studies have produced good outcomes in reducing anger and aggression and improvements have been maintained for assessment. Many of these case studies (e.g. Black & Novaco, 1993; Murphy & Clare; 1991; Rose & West, 1999) have been conducted in hospital and community settings with individuals who have been assessed as having histories of aggressive behaviour.

  There have also been a small number of case studies of cognitive behavioural anger treatment involving offenders with ID that have yielded positive outcomes. Lindsay, Allan, Macleod, Smart, and Smith (2003) reported on six men with mild ID all of whom had been involved with the criminal justice system (CJS) for reasons of aggression and violence. Anger treatment involved group therapy that included: psycho education, arousal reduction, role-play, problem solving, and stress inoculation through imagination of anger-provoking situations. Treatment was assessed using self-report measures, reports of aggressive incidents and an anger diary. All measures improved significantly and these improvements were maintained at 15-month follow up. None of the participants had been violent at four-year follow-up.

  PHOTO 21.2 The purpose of treatment is to equip ID offenders with thinking skills that will promote alternative, prosocial means of approaching situations in which the person is at risk of offending.

  Source: © Brad Killer/Getty Images

  In another single case study, Novaco and Taylor (2005) described individual Anger Management Training for a man who was referred for violent behaviour in a medium secure setting. Self-report for anger and response to provocation was significantl
y reduced (by 3.3 and 2.8 standard deviations respectively). Care staff observations of reductions in violence were convergent with these findings and improvements were maintained to four-month follow up. Allan, Lindsay, Macleod, and Smith (2001) report treatment for five women with mild to borderline ID, who all had been involved with the CJS for reasons of violence. Treatment was similar to that described by Lindsay et al. (2003), with significant improvements following intervention and at 15-month follow up, Only one of the women had a further incident of violence and improved scores on the self-reported inventories were maintained. While these case evaluations have generally demonstrated a positive outcome, they have all been uncontrolled demonstrations of effectiveness.

  21.5.2 Controlled Studies of Anger Management Training (AMT)

  There have now been a number of controlled trials of AMT. In a randomised controlled trial of anger management, Willner, Jones, Tams, and Green (2002) invited community support teams to refer individuals who would benefit from attendance at an anger management group. They then allocated 14 individuals randomly to the treatment and control conditions. Treatment consisted of nine two-hour sessions but not all individuals attended every group session. The intervention was evaluated using two inventories of anger-provoking situations completed independently by clients their carers. Clients in the treated group improved on both self and carer ratings relative to their own pre-treatment scores and to the control post-treatment scores. Clients in the treated group showed further improvement at three-month follow-up. Although this study did not recruit individuals through the criminal justice system, it did use a randomised control procedure and as a result has strengths not evident in waiting list controlled comparisons.