I need help with this assignment!!!! I have included both articles that this assignment must come from NO OTHER SCOLARY REFERENCES PLEASE!!!!! I have to utlize these two artcles only. I have also incl

An Analysis of Dangerous Sexual Oﬀender Assessment Reports:

Recommendations for Best Practice Dominic J. Doyle*, James R. P. Ogloﬀ and Stuart D. M. Thomas Centre for Forensic Behavioural Science, School of Psychology and Psychiatry, Monash University and Victorian Institute of Forensic Mental Health (Forensicare), Clifton Hill, Victoria, Australia The operation of Australia’s preventive detention legislation depends upon forensic clinician assessments of risk for future sexual oﬀending. However, to date, no information is available regarding how such assessments of risk are being conducted.

This study provides the ﬁrst descriptive analysis of the risk assessment practices of mental health professionals conducting assessments under preventive detention legislation around Australia. Eighty-six forensic evaluation reports on 56 sex oﬀenders subject to preventive detention proceedings were obtained and analysed. Overall, the ﬁndings are mixed. Positively, valid structured risk assessment tools were commonly utilized. Also, there was good agreement between experts on the ﬁnal risk assessment outcome, suggesting a consensus in relevant areas relating to risk assessment. However, a number of concerning results were also found (e.g., some evaluators adopted invalid risk assessment methodologies; others incorrectly applied and interpreted otherwise valid risk tools). Taken together, the ﬁndings suggest that the standard of practice of risk assessment must be raised. Recommendations for best practice are proposed.

Key words:best practice; legislation; preventive detention; risk assessment; sex oﬀenders.

‘‘Better get a lawyer son. Better get a real good one.’’ (Cruel Sea, 1995) The assessment of risk for future violence is central to many decisions made within the criminal justice system (McSherry & Key- zer, 2009; McSherry, Keyzer, & Freiberg, 2006). Decisions with regard to bail appli- cations, sentencing, parole, and conditions of release from custody may all be aﬀected by the oﬀender’s perceived level of risk for violence (Glazebrook, 2010). Accordingly, the law often turns to clinicians – particu- larly psychiatrists and psychologists – for opinions on the level of risk for violenceposed by an oﬀender (Ogloﬀ & Davis, 2005). Recently, however, clinician opi- nions of risk for future violence have been aﬀorded an even greater responsibility following the enactment of unique legisla- tion targeting sexual oﬀenders.

In recent years, a number of Australian states have introduced laws allowing for the continued detention or community supervision of sex oﬀenders whose sen- tences have expired but who are still considered to be dangerous (Doyle & Ogloﬀ, 2009; Sentencing Advisory Council, 2006). The enactment of these preventive detention measures continues an in- ternational trend in the proliferation of Correspondence: Dominic J Doyle, Centre for Forensic Behavioural Science, Monash University, 505 Hoddle Street, Clifton Hill Victoria 3068, Australia. Email: [email protected] Psychiatry, Psychology and Law Vol. 18, No. 4, November 2011, 537–556 ISSN 1321-8719 print/ISSN 1934-1687 online 2011 The Australian and New Zealand Association of Psychiatry, Psychology and Law http://dx.doi.org/10.1080/13218719.2010.499159 http://www.tandfonline.com legislation designed to reduce risks of sexual recidivism (Mercado & Ogloﬀ, 2007; Vess, 2009b). Most recently, the government in Victoria has introduced legislation expanding the scope of the post-sentence community supervision pro- visions for sex oﬀenders to allow for their ongoing detention in prison (Victoria, Legislative Assembly, 2005).

In contrast to traditional criminal justice principles, preventive detention leg- islation is predicated not on the crimes previously committed by an oﬀender and tied to a ﬁnding of guilt, but rather on the oﬀender’sriskto commit other sexual crimes in the future. This shift in focus from previous oﬀending to the risk of future oﬀending has placed the clinical risk assessment as the paramount consideration in preventive detention proceedings (Gla- zebrook, 2010). Indeed, under the legisla- tion, psychologists and psychiatrists are required to prepare reports that assess the level of risk or likelihood that the oﬀender would commit further sexual oﬀences if not detained in prison or supervised for an extended period upon release. Moreover, the courts are statutorily required to take into account this clinical assessment of risk in deciding whether to preventively detain or supervise the nominated oﬀender (Doyle & Ogloﬀ, 2009). While the experts’ reports are not necessarily dispositive (Freckelton & Selby, 2009), very often the court’s decision of whether to impose the order turns critically upon the clinician’s opinion of risk for future sexual violence (Ogloﬀ & Doyle, 2009).

The law’s dependence on risk assess- ment for the operation of this legislation places a considerable burden on the clin- ician and raises expectations that are perhaps impossible to attain. Accordingly, concerns held by mental health profes- sionals regarding the validity and precision of risk assessment approaches and technol- ogies have intensiﬁed (Ogloﬀ & Doyle, 2009), and, in turn, raised doubts about theappropriateness of this legislation (Doyle & Ogloﬀ, 2009).

However, irrespective of these misgiv- ings, numerous preventive detention pro- ceedings have occurred nationally and hundreds of risk assessment reports have been tendered to the courts. Yet, to date, nothing is known about how clinicians go about the task of assessing risk for future sexual violence in Australia for no sys- tematic information has heretofore been analysed and published.

The current study presents the ﬁndings of a descriptive analysis of risk assessment reports prepared by mental health profes- sionals pursuant to Australia’s preventive detention legislation. It is vitally important to establish an understanding of clinicians’ risk assessment practices within this high- stakes legal context to ensure that legal decision-makers are provided with the highest quality of expert opinion on risk and to preserve and reinforce professional standards.

This article will ﬁrst provide an over- view of current approaches to risk assess- ment. While a comprehensive review of the risk assessment literature is beyond the scope of this article, some contemporary approaches will be highlighted. Second, some of the theoretical and practical issues that limit the precision of risk assessment will be outlined. Finally, the descriptive analysis of Australian forensic clinicians’ dangerous sexual oﬀender assessment re- ports will be presented. A number of recommendations for best practice in the assessment of risk for future sexual vio- lence will be proposed.

Contemporary Approaches to Sex Oﬀender Risk Assessment Within the last 15 years, substantial re- search eﬀorts to develop and enhance risk assessment technologies have resulted in the development of numerous formal tools for assessing risk for future sexual violence 538D. J. Doyle et al. (Douglas & Skeem, 2005; Ogloﬀ & Daﬀern, 2004). These can be divided into two broad camps: actuarial models and structured professional judgement (SPJ).

Actuarial prediction Actuarial tools comprise variables that have been found to have a statistical relationship to subsequent oﬀending (Ogl- oﬀ & Davis, 2005). The ﬁnal actuarial model consists of the combination of risk factors that demonstrated the strongest statistical relationship to the predicted out- come (i.e., sexual oﬀending). The Static-99 (Hanson & Thornton, 1999), a 10-item in- strument, is one of the most popular actuarial tools for the prediction of future sexual oﬀending (Hanson, Morton, & Har- ris, 2003). Numerous validation studies, and, more recently, a meta-analysis of the accu- racy of risk assessment instruments, demon- strate that this instrument reliably provides assessments of risk with a moderate degree of accuracy (Barbaree, Seto, Langton, & Pea- cock, 2001; Hanson & Morton-Bourgon, 2007; La˚ ngstro ¨m, 2004).

Structured professional judgement SPJ instruments consist of empirically informed professional guidelines to assist clinicians to develop an assessment of risk (Hart, Kropp, & Laws, 2003). Similar to actuarial prediction, SPJ tools also consist of risk factors derived empirically and rationally from the research literature.

However, in contrast to the actuarial model, rather than summing the items in a mechanical fashion, clinicians formulate a structured clinical opinion of low, mod- erate, or high risk (Davis & Ogloﬀ, 2008).

The SPJ approach takes into account both historical and dynamic risk factors, and allows clinicians to utilize their professional judgement within a structured framework, so that idiosyncratic butimportant characteristics of the individual that pertain to risk are considered.

The Sexual Violence Risk-20 (SVR-20; Boer, Hart, Kropp, & Webster, 1997) and the Risk for Sexual Violence Protocol (RSVP; Hart et al., 2003) are examples of sexual risk instruments based on the SPJ model. Due to their somewhat recent development, the SPJ approach has only been evaluated in a handful of studies; although this research has generally been quite promising (e.g., Craig, Browne, & Stringer, 2004; de Vogel, de Ruiter, van Beek, & Mead, 2004; Hanson & Morton- Bourgon, 2007; Macpherson, 2003).

Despite the often heated debate in the literature regarding the relative merits and predictive superiority of actuarial and SPJ methods (see Harris, Rice, & Quinsey, 2008; Hart, Michie, & Cooke, 2007; Quin- sey, Harris, Rice, & Cormier, 2006), both approaches have comparable predictive validity (Hanson & Morton-Bourgon, 2007).

As noted, the ﬁeld of risk assessment has advanced considerably in recent years.

Indeed, the eﬀect size for violence risk assessment is now superior to that of many other medical and psychological practices (Davis & Ogloﬀ, 2008). Nevertheless, despite these advances, the assessment of risk is a complex task and there remain theoretical and practical limitations on eﬀective prediction in the individual case (Mullen & Ogloﬀ, 2009).

A Cautionary Tale A full critique of the issues that limit the reliability and validity of risk assessment is beyond the scope of this article (see Ogloﬀ & Doyle, 2009). However, some salient concerns will be brieﬂy reviewed; for the limitations of risk assessment are as relevant as the very outcome of the assessment itself. From the outset, the practical issue of the base rate of sexual reoﬀending serves to curtail the precision An Analysis of Australian Forensic Reports539 of risk assessment (Wollert, 2006). As explained in greater detail elsewhere (see Mullen & Ogloﬀ, 2009; Ogloﬀ & Davis, 2005), the less common the future beha- viour under prediction in the population, the less accurate the predictions. And, contrary to the popular opinion, sexual reoﬀending is not a high-frequency occur- rence (Doyle & Ogloﬀ, 2009). Indeed, sexual recidivism research consistently ﬁnds that, as a group, most sex oﬀenders do not go on to sexually reoﬀend (e.g., a meta-analysis of sexual recidivism includ- ing 30,000 sex oﬀenders found an average recidivism rate of 13.7% over 5–7 years) (Hanson & Bussie` re, 1998; Hanson & Morton-Bourgon, 2005). Given this rela- tively low base rate of sexual recidivism and the practical diﬃculties in predicting a low base rate event, attempting to predict who will commit further serious sexual oﬀences will inevitably be accompanied by false accusations (Doyle & Ogloﬀ, 2009).

A signiﬁcant issue that compromises the validity of actuarial instruments con- cerns the unreliability of applying the group-based risk evaluation of an actuarial tool to the assessment of risk in the individual case (Berlin, Galbreath, Geary, & McGlone, 2003; Hart et al., 2007; Mullen & Ogloﬀ, 2009). For instance, if an oﬀender scores 6 on the Static-99 instrument he is considered to be in the ‘‘high-risk’’ category, 52% of whom (in the original sample) were known to reoﬀend throughout a 15-year follow-up. However, the instrument cannot specify whether the ‘‘high’’ risk oﬀender belongs to the 52% of people in this category who sexually reoﬀended, or to the 48% of people who did not (Berlin et al., 2003). Therefore, an individual’s score on the actuarial tool fails to be a reliable guide to the individual’s speciﬁc risk to sexually reoﬀend, for the simple reason that actuarial methods are not designed to assign levels of risk to individuals but to groups (Mullen & Ogl- oﬀ, 2009).Other considerations that arguably weaken the validity of actuarial models of risk assessment have been detailed else- where (see Doyle & Ogloﬀ, 2009; Ogloﬀ & Doyle, 2009), and include the tools’ in- sensitivity to the impact of age on risk, the lack of parallel between the legal question to be answered and the evaluative results of the instruments, and the limited data validating the instruments’ use locally.

Importantly, however, these limitations also emphasize that when mental health professionals are asked to provide opinions of risk for future violence to courts, or other decision-making bodies, it is incum- bent upon the clinician to be clear to state the limitations to the science upon which their ﬁndings rest (Mullen & Ogloﬀ, 2009).

The Present Study The number of publications concerned with the role of risk assessment under preventive detention legislation is consider- able and continues to grow (i.e., Keyzer, Pereira, & Southwood, 2004; McSherry, 2005; McSherry et al., 2006; Mercado & Ogloﬀ, 2007; Mullen & Ogloﬀ, 2009; Ogl- oﬀ & Doyle, 2009; Scott, 2008; Sentencing Advisory Council, 2006; Smallbone & Ransley, 2005; Vess, 2009a, 2009b; Wood & Ogloﬀ, 2006). Despite this attention being paid to the issue, to date, there has been no empirical evaluation regarding howsuch assessments of risk are being conducted under these laws.

Commonly, investigations of clinical practice are achieved by surveying the professionals (e.g., Allan, Martin, & Allan, 2000; Martin, Allan, & Allan, 2001; Mer- cado, Elbogen, Scalora, & Tomkins, 2001).

However, a limitation to this methodology is its vulnerability to self-report biases. A more objective assessment of how forensic clinicians go about the task of risk assess- ment is achieved via an analysis of the actual reports that they tender to the court.

While content analyses of reports produced 540D. J. Doyle et al. under equivalent legislation in the United States have been conducted (see Amenta, 2005; Levenson, 2004), the present study represents the ﬁrst analysis of reports prepared by mental health professionals conducting assessments of risk for sexual recidivism under Australia’s preventive detention legislation. This research seeks to develop a greater understanding of the methodology and standard of practice among forensic clinicians providing expert evidence to the courts within this particular legal context.

We believe this investigation is impor- tant for a number of reasons. First, this research will enable a measure of the quality of expert opinion on risk being provided to legal decision-makers. Second, if it is found that the practice of risk assessment is inadequate then this will need to be immediately remedied given that compromised risk assessments have sig- niﬁcant implications for public safety, the civil liberties of oﬀenders, and the integ- rity of the professions to which the evaluators are ascribed. In light of these considerations, the value of an analysis of the state of forensic practice in this legal area is clear.

Method Sample Eighty-six forensic evaluation reports on 56 sex oﬀenders subject to preventive detention legal proceedings were made available to the researchers. The sample consisted of 27 (31.4%) reports from Victoria, 33 (38.4%) reports from Western Australia, and 26 (30.2%) reports from New South Wales. The reports ranged in date from May 30, 2005 to February 2, 2009. Consistent with statutory language all subjects in New South Wales and Western Australia were examined by two forensic evaluators, though at the time of data collection in three cases only one psychiatric report was made availableto the researchers. In Victoria two appli- cations were accompanied by two assess- ment reports. In summary, the researchers were provided access to 60 reports produced in relation to 30 oﬀen- ders and 26 reports written in relation to 26 oﬀenders.

Report Coding Procedure Data were recorded via review of the aforementioned forensic evaluation re- ports. Based upon a comprehensive review of the literature, the authors developed a coding manual detailing the scoring criteria for variables of interest. The manual was modelled on two other coding instruments used in previous report analyses (Amenta, 2005; Skeem, Golding, Cohn, & Berge, 1998) and underwent several iterations and a pilot analysis. This analysis resulted in a reﬁning of the coding instrument to ensure that consistency of information was ob- tained across reports (a copy of the data collection form is available from the authors upon request).

Data Collection Procedure To obtain access to the forensic reports, written requests were sent to the Chief Justices of the Supreme Courts of Queens- land, Western Australia, and New South Wales. In Victoria, although both the Supreme and County Courts are eligible to hear post-sentence proceedings, the lower court has heard the vast majority of post-sentence applications. Accordingly, a request was only sent to the Chief Judge of the County Court of Victoria. Although authorities in Queensland declined partici- pation, all other states consented. Upon receipt of the reports, raw, de-identiﬁed data were transcribed onto the coding manual. Unique identiﬁers were assigned to the oﬀender, the reports, and the evaluator. Full ethical approval from Monash University was received. An Analysis of Australian Forensic Reports541 The reports typically included psycho- social, criminal history, diagnosis, and risk assessment information. Speciﬁcally, the information of interest to the researchers were organized around three central themes. First, the rater coded (a) the exami- ners’ general assessment practices (e.g., number and length of interviews), (b) whether the examiner included the referral source and purpose of the assessment, and (c) whether the examiner described notify- ing the defendant about the purpose and conﬁdentiality of the evaluation and docu- mented their consent to proceed.

Second, the rater coded (a) the types and number of risk assessment methodol- ogies and instruments employed, (b) the manner in which risk assessment results were communicated, (c) the method of communicating the ﬁnal opinion of risk, (d) the nature of any statements of limita- tions pertaining to the practice of risk assessment, and (e) if examiners identiﬁed factors associated with risk outside of a structured assessment, whether such fac- tors have either robust or equivocal sup- port in the literature on sexual recidivism.

Third, for the purposes of assessing inter-rater reliability, the rater coded (a) Axis I and Axis II diagnoses, (b) risk scores on the Static-99 and PCL-R, and (c) the ﬁnal risk rating provided by each examiner.

Analyses Similar to previous analyses of psychiatric and psychological reports both domesti- cally (Allan et al., 2000; Martin et al., 2001) and internationally (Amenta, 2005; Heil- brun & Collins, 1995; Heilbrun, O’Neill, Strohman, Bowman, & Philipson, 2000; Larkin & Collins, 1989; Petrella & Poy- thress, 1983; Skeem et al., 1998), the ﬁndings of this study were predominantly descriptive. As the ﬁrst empirical analysis of clinicians’ risk assessment practices under Australia’s preventive detention leg- islation, this was deemed appropriate.Inter-rater reliability was examined on the total score of the Static-99 using the intra-class correlation coeﬃcient (ICC) two-way model for continuous variables (Bartko, 1966; Shrout & Fleiss, 1979). The PCL-R was inconsistently reported as a numerical value, category, or percentile between evaluators. Therefore, all scores and percentiles were transformed into a categorical rating of low (less than 20), medium (between 20 and 30), or high (more than 30). Thus, the PCL-R inter- rater reliability was assessed using kappa coeﬃcients.

DSM-IV (American Psychiatric Asso- ciation, 2000) diagnoses and ﬁnal risk rating were dichotomous variables and were also assessed using kappa coeﬃcients.

Kappa coeﬃcients were computed when a 262 table was attained. Given the rela- tively small number of cases included in the reliability analyses, and the disproportion- ate impact on kappa values this can have, levels of agreement (%) are also provided.

While diﬀerent interpretations of reliability coeﬃcients exist, this study, consistent with Levenson’s (2004) approach, adopted a higher standard given the seriousness of the decisions being made in this legal context. For this study, a reliability coeﬃ- cient below .60 is considered poor, .60 to .74 is considered fair, and .75 to 1.0 is considered good (Bloom, Fischer, & Orme, 1999).

Results Participants Report authors Twenty-three mental health professionals authored 86 reports. Sixteen psychiatrists authored 60 (69.8%) reports, with 14 (87.5%) psychiatrists indicating that they had a specialization in the forensic ﬁeld.

Seven psychologists authored 26 (30.2%) reports. Of those with psychology training, four (57.1%) evaluators had received post- graduate qualiﬁcations (i.e., Doctor of 542D. J. Doyle et al. Psychology), while the highest level of qualiﬁcation for three 1(42.9%) evaluators was Honours or Graduate Diploma in Psychology. The number of reports per evaluator ranged from 1 to 9. Sixteen evaluators (70%) had ﬁve or fewer reports included in the sample and seven evaluators (30%) had between ﬁve and nine reports.

Oﬀenders under evaluation The demographic, clinical, and criminal characteristics of those subject to these forensic evaluations have been reported elsewhere (Doyle, Ogloﬀ, & Thomas, in press). Brieﬂy, their mean age was 44.7 years (SD¼14.2), and 10 (17.9%) were known to be in a relationship at the time of the legal proceedings. The majority (n¼45, 80.4%) had less than a high school education. Almost two thirds (n¼35, 62.5%) were reported to have had a substance abuse problem throughout their lifetime, and more than two thirds (n¼38, 67.9%) were currently diagnosed with an Axis I disorder. Thirty-ﬁve (62.5%) received a paraphilia diagnosis, the most common of which was a diagnosis of paedophilia (n¼29, 51.8%). More than half (n¼30, 53.6%) were diagnosed with a personality disorder, with antisocial per- sonality disorder (n¼20, 35.7%) being the most common.

The vast majority of the oﬀenders (n¼51, 91.1%) had previous convictions for sexual oﬀences prior to their index oﬀence, with a mean number of prior sentencing dates of 2.91 (SD¼2.7). There mean age at conviction for ﬁrst sex oﬀence was 23.8 years (SD¼11.0), with over one third (n¼20, 35.7%) committing their ﬁrst sexual oﬀence prior to the age of 18 years. Twenty-ﬁve (44.6%) oﬀenders had a history of oﬀending against male victims, while the majority (n¼47, 83.9%) had a history of sexual oﬀences against female victims. Most oﬀenders had a history of oﬀending outside the family (n¼53,94.6%) and a history of violent sexual oﬀending (n¼33, 58.9%). The majority (n¼51, 91.1%) also had prior convictions for non-sexual oﬀences.

Report Writing Characteristics General assessment practices All reports indicated the number of inter- views conducted with the oﬀender; the average was 1.62 (SD¼0.85) but ranged from 1 to 5. A majority of reports (n¼79, 91.9%) were based on either one or two interviews. The total length of the inter- views was noted in 67 (77.9%) reports, with a mean length of 234.81 minutes (3.9 hrs) (SD¼129.24), ranging from 90 min- utes (1.5 hrs) to 645 minutes (10.75 hrs).

Inclusion of information Almost two thirds (n¼56, 65.1%) of reports included a statement that identiﬁed the authority that requested the evaluation, while 30 reports (34.9%) omitted this information. The purpose of the assess- ment was clearly articulated in 17 (19.8%) reports. The majority (n¼66, 76.7%) indirectly referred to the reason for assess- ment by reference to the legislation, while 3 (3.5%) reports omitted any reference to the purpose of the assessment. All reports indicated that the author had engaged in a review of collateral information.

Documentation of notiﬁcation Less than two thirds of the reports (n¼53, 61.6%) included a statement that the oﬀender was notiﬁed regarding the limits to conﬁdentiality. Consent to participate in the assessment was documented in 51 (59.3%) reports, while 39 (45.3%) reports included a statement that the oﬀender was told the nature and purpose of the evalua- tion. That the oﬀender understood the information contained within the notiﬁca- tion was documented in 46 (53.5%) reports. An Analysis of Australian Forensic Reports543 Risk Assessment Practices Risk assessment methods Table 1 presents the type, frequency, and combination of risk assessment methods employed by the evaluators. Multiple meth- ods of risk assessment were regularly uti- lized, with 2 (n¼35, 40.7%) and 3 (n¼35, 40.7%) methods being the most common.

The methods of unstructured clinical judge- ment and adjusted actuarial, were, respec- tively, used in 18 (21%) and 24 (27.9%) reports. The most common combination of methods comprised actuarial and SPJ (n¼25, 29.1%).

Risk assessment tools A range of risk assessment instruments derived from actuarial, adjusted, and SPJ methodologies were employed by the eva- luators (see Table 2). The Static-99 was clearly the most frequently used riskassessment tool. The SPJ tools of the SVR- 20 and RSVP were utilized in 44 (51.1%) reports.

Reporting and interpreting the Static-99 Table 3 presents frequency data across a number of dimensions relevant to reporting Static-99 results and interpreting the tool’s probability estimates. Of those Table 1. Frequency and type of risk assess- ment methods employed by evaluators.

Risk assessment methodN% Actuarial Alone 3 3.5 ActuarialþDynamic a 22.3 ActuarialþSPJ b 25 29.1 ActuarialþDynamicþSPJ 5 5.8 ActuarialþAdjustedþSPJ 4 4.7 Clinical Judgement Alone 5 5.8 Clinical JudgementþActuarial 8 9.3 Clinical JudgementþActuarialþ Dynamic11.2 Clinical JudgementþActuarialþ Empirically Guided11.2 Clinical JudgementþActuarialþ AdjustedþSPJ33.5 Empirically Guided Alone 1 1.2 Empirically GuidedþActuarial 4 4.7 Empirically GuidedþActuarialþ Adjusted17 19.8 Empirically Guidedþ ActuarialþSPJ78.1 aDynamic refers to tools that, though consisted of dynamic variables, were not used to adjust the assessment of risk based on historical factors. bSPJ is an abbreviation for Structured Professional Judgement. Table 2. Frequency and type of risk assess- ment tools employed by evaluators.

Risk assessment toolN% Static-99 79 91.9 RSVP 31 36.0 SONAR 23 26.7 SVR-20 13 15.1 3-Predictor Model 9 10.5 RRASOR 4 4.7 SORAG 1 1.2 HCR-20 12 14.0 PCL-R a 46 53.5 aAlthough not designed to be a risk assessment tool, the PCL-R has been reliably associated with both violent and sexual recidivism.

Table 3. Reporting and interpreting Static-99 results.

Components of Static-99 reporting N a %b Inclusion of Static-99 results Total Static-99 score 52 65.8 Probability estimate 66 83.5 Base rate data 7 10.6 Interpretation of the probability estimate Group-based risk 38 57.6 Individual’s risk 10 15.2 Unclear/contradictory 15 27.3 Errors in reporting probability estimates Incorrect recidivism percentages14 21.2 Uncollapsed recidivism percentages57.6 Impact of oﬀender’s age on Static-99 result15 19 aNumber of reports that included the described information. bRefers to the percentage of reports that provided the relevant information relative to those reports that used the Static-99 tool. 544D. J. Doyle et al. reports that used the Static-99 risk assessment, the majority included the probability estimate (n¼66, 83.5%). A number of reports contained errors in the reporting of the evaluative results (n¼19, 28.8%).

Stating the limitations of risk assessment The type and frequency of statements provided by assessors regarding the limita- tions of the Static-99 risk tool speciﬁcally, and the practice of risk assessment gener- ally are listed in Table 4.

Static-99 speciﬁc statements of limita- tions were provided in 46 (58.2%) re- ports, ranging from 0 to 5. The limitations most commonly stated con- cerned the diﬃculties in applying the group estimate of risk to the individual case (n¼36, 45.6%).

A general statement of the limits to the practice of risk assessment was provided in 39 (45.3%) reports, ranging from 0 to 3. The statement mostcommonly provided concerned the limited accuracy of risk assessment (n¼29, 33.7%).

Risk factors In 52 (60.5%) reports, evaluators identiﬁed factors outside of a formal (i.e., instru- ment-based or empirically guided) risk assessment procedure, that they considered to be associated with an elevated risk for reoﬀence. Table 5 lists those risk factors identiﬁed by evaluators that do have empirical support for being associated with recidivism risk. Prior sex oﬀences (n¼27, 31.4%), deviant sexual pre- ferences (n¼31, 36.0%) and lack of social/familial/community support (n¼22, 25.6%), were the individual risk factors most commonly identiﬁed by evaluators and supported by the sexual recidivism literature to be associated with risk of sexual reoﬀending.

Additionally, factors that were identi- ﬁed by evaluators external to a formal risk assessment procedure as being associated with risk, but that do not have strong empirical support, are presented in Table 6.

Thirteen (56.5%) evaluators wrote 28 (32.6%) reports within which they identi- ﬁed such ‘‘risk’’ factors. Minimizing culp- ability (n¼17, 19.7%), and denial (n¼9, 10.5%) were the two factors most com- monly believed to be associated with sexual recidivism that are currently lacking in empirical support.

Communication of ﬁnal risk rating A ﬁnal opinion of risk was provided in 79 (91.9%) reports. Of those reports to include a ﬁnal opinion, all but one (n¼78, 98.7%) utilized a categorical method of risk communication (i.e., high, moderate, low). The majority of assess- ments concluded the oﬀender posed a high risk (n¼64, 74.4%) of sexual reoﬀending.

Five reports (5.8%) provided a risk rating Table 4. Frequency of statements of limita- tions provided for Static-99 and general risk assessment.

Statement of limitation N % Static-99 risk assessment Moving from group to individual estimations of risk36 45.6 Absence of dynamic factors 23 29.1 Not validated on Australian sex oﬀenders18 22.8 Accuracy of the instrument 13 16.5 Ethical issues regarding its use in the legal context11 13.9 Dissimilar deﬁnition of sex oﬀence between Static-99 and legislation67.6 Not validated on indigenous sex oﬀenders33.8 General risk assessment Accuracy of risk assessment 29 33.7 Limitations to the science of risk prediction18 20.9 Not entirely objective process 9 10.5 Tools not validated for use in Australia22.3 An Analysis of Australian Forensic Reports545 of ‘‘very high.’’ Risk ratings of moderate- high, moderate, and moderate-low were, respectively, noted in 5 (5.8%), 2 (2.3%), and 3 (3.5%) reports. A risk rating of ‘‘likely’’ was provided in one report (1.2%).

In addition to describing the subject as ‘‘high risk’’ for future sexual oﬀending, 16 (18.5%) reports also described the oﬀender as being ‘‘some risk,’’ a ‘‘signiﬁcant risk,’’ a ‘‘virtually certain risk,’’ and ‘‘very,’’‘‘signiﬁcantly,’’ and an ‘‘unacceptably’’ high risk.

Inter-Rater Reliability Risk scores, diagnoses, and ﬁnal risk ratings All inter-rater reliability results are dis- played in Table 7. Analysis of the Static-99 produced an ICC coeﬃcient of .85. While the level of agreement was moderate (65%), the inter-rater reliability (Pearson r¼.81) was high. Static-99 scores diﬀered in seven cases; in all but one the diﬀerence in Static-99 scores did not correspond with a diﬀerence in the associated risk rating.

The PCL-R was used by both evaluators in only 10 cases with good levels of agreement (70%), but poor reliability (k¼.46).

Reliability of Axis I diagnoses ranged from good to excellent. Paedophilia de- monstrated excellent reliability (k¼.93), as did psychotic disorders (k¼1.0). The reliability of Axis II diagnoses were poor, except for psychopathy (k¼1.0).

The level of agreement between evalua- tors on the ﬁnal risk rating was very good (84.7%).

Table 5. Frequency of empirically supported risk factors.

Factors N % Demographic factors44.7 Age 1 1.2 Martial status 2 2.3 Employment history 3 3.5 Criminal history22 25.6 Total number of prior oﬀences89.3 History of rule violation 13 15.1 Sexual criminal history29 33.7 Prior sex oﬀences 27 31.4 Stranger victims 6 7.0 Extrafamilial victims 9 10.5 Early onset of sexual oﬀending11.2 Male victims 11 12.8 Diverse sexual crimes 10 11.6 Sexual deviance32 37.2 Deviant sexual preferences 31 36.0 Sexual preoccupations 5 5.8 Personality disorder Psychopathy 2 2.3 Antisocial personality disorder 6 7.0 Any personality disorder 1 1.2 Treatment history78.1 Failure to complete treatment 3 3.5 Failure to participate in treatment44.7 Dynamic factors41 47.7 Sexual attitudes tolerant of sexual violence78.1 Intimacy deﬁcits 14 16.3 Lack of appropriate adult sexual relationship11 12.8 Impulsivity 12 14.0 Substance abuse 14 16.3 Circumstances post release (e.g., release plans)89.3 Lack of social/familial/ community support22 25.6 Psychological problems (i.e., negative mood)55.8 Table 6. Frequency of identiﬁed risk factors with equivocal support.

FactorsN%d a Minimizing culpability 17 19.7 .00 Denial 9 10.57.02 Low treatment motivation 8 9.37.02 Victim empathy 7 8.17.01 Low intelligence 7 7.0 .04 Prior history of violent oﬀending6 7.0 .01 Victim of sexual abuse 3 3.5 .02 Degree of force used 2 2.3 .00 Degree of sexual contact 1 1.27.16 Adverse childhood environment1 1.2 .00 aThe standardized mean diﬀerence statistic is taken from Hanson and Morton-Bourgon’s (2004) updated meta-analysis of predictors of sexual recidivism.

According to Cohen (1988), d values of .20 are considered small. The value of d is approximately twice as large as the correlation coeﬃcient calculated from the same data. 546D. J. Doyle et al. Discussion The assessment of an oﬀender’s risk for further sexual oﬀending is central to preventive detention proceedings. This in- vestigation provided the ﬁrst descriptive analysis of forensic evaluators’ risk assess- ment practices and the reliability of risk assessment outcomes in these legal matters.

Despite the descriptive nature of this study, some important conclusions can be drawn, and practical recommendations made.

Taken together, the ﬁndings of this investigation are mixed. That the majority of clinicians employed valid structured tools to assess future SVR is encouraging, and indicates a signiﬁcant translation of empirical research into clinical practice.

Encouraging too was that there was good agreement between the experts on the ﬁnal risk assessment outcomes, suggesting a consensus in relevant areas relating to risk assessment. Despite these relatively positive ﬁndings, however, a number of disconcert- ing results were also found. For example, some evaluators adopted invalid risk as- sessment methodologies. Others incorrectlyapplied and interpreted otherwise valid risk tools. Also, the limits that constrain the science of risk assessment were all too infrequently communicated. Given that these legal proceedings involve fundamen- tal questions of individual liberty and public safety, these are egregious errors.

Overall, the ﬁndings suggest that the standard of practice of risk assessment must be raised. In what follows the results across the domains of interest will be discussed. Recommendations for best prac- tice will be proposed.

Report Writing and Assessment Practices Under the code of ethics applicable to the professions of psychology and psychiatry (Australian Psychological Society, 2007; Royal Australian and New Zealand Col- lege of Psychiatrists, 1998), practitioners are obligated to provide the examinee with a notiﬁcation outlining the nature and purpose of an assessment, the limits to conﬁdentiality that pertain to the assess- ment, and obtain the person’s informed Table 7. Inter-rater reliability coeﬃcients.

Measures and variablesNAgreement % ICC Kappa a Risk assessment instruments Static-99 20 65.0 .81 PCL-R 10 70.0 .46 Diagnoses Paedophilia 30 96.7 .93 Other paraphilias 30 96.7 Multiple paraphilias 30 93.3 .76 Psychotic disorder 30 100.0 1.0 Other Axis I disorder 30 96.7 .78 Antisocial PD 30 83.3 .65 Psychopathy 30 100.0 1.0 Personality disorder NOS 30 73.3 .19 Other personality disorder 30 86.77.053 Final risk rating24 84.7 High b 24 79.2 .42 Moderate c 24 79.2 .32 Moderate low 95.8 aKappa values were only available when a 262 table could be attained. bVery high and high ratings were merged to form a single rating of high for the purposes of cross-tabulation. cModerate-high and moderate ratings were merged. An Analysis of Australian Forensic Reports547 consent to proceed with the interview.

However, a signiﬁcant proportion of re- ports failed to document that the various constituents of the notiﬁcation had taken place. We do not presume that failure to document the notiﬁcation equates to fail- ure to provide the notiﬁcation. Never- theless, the careful documentation of the notiﬁcation is advised so that the fulﬁlment of the ethical obligation to notify is formally recorded and the assessor is protected from claims to the contrary.

Contrary to general principles of for- ensic report writing (e.g., Allnutt & Cha- plow, 2000), a number of reports failed to identify the authority (e.g., Supreme Court) that requested the evaluation and clearly articulate the reason for the referral.

For clarity, the authors recommend that the referrer and the purpose of the assess- ment be clearly documented.

Risk Assessment Practices Methods of risk assessment Despite ongoing debate among experts regarding the relative merits of various sex oﬀender risk assessment methods, some broad points of agreement are being reached. These emerging points of agree- ment are that empirically validated actuar- ial measures best form the foundation of risk assessment while a structured consid- eration of dynamic risk factors assist in formulating the nature of the risk pre- sented by the oﬀender and a management strategy to reduce such risks (Vess, 2009b).

A signiﬁcant number of reports ap- proached the task of risk assessment in this way, combining actuarial and SPJ methods. It is positive that their expert opinion has been grounded in the best risk assessment methods available.

Less encouraging was the ﬁnding that a number of clinicians utilized an unstruc- tured clinical judgement approach in their assessment of SVR. Simply, the empirical evidence does not support unaided clinicaljudgement as a valid method of risk assess- ment (Hanson & Morton-Bourgon, 2007; Ogloﬀ & Davis, 2005). Some may advocate an unstructured approach to risk assessment is necessary when no relevant structured tools are available (e.g., when required to assess risk for sexual recidivism in women oﬀenders). However, in such cases experts are cautioned from providing an opinion that is without empirical foundation.

Several reports, problematically, pre- sented an opinion on risk based solely upon the results of an actuarial method. The actuarial approach provides a valid, yet incomplete assessment of risk. Even the instrument’s authors advise that the Static- 99 is not comprehensive because it ‘‘ne- glects whole categories of potentially rele- vant variables’’(Hanson & Thornton, 1999, p. 18). Clinicians are cautioned from relying exclusively upon an actuarial method at the expense of a more comprehensive, multi- modal risk assessment procedure.

A signiﬁcant number of reports utilized the adjusted-actuarial method as part of their risk assessment. While the considera- tion of dynamic risk variables is relevant to risk assessment (Douglas & Skeem, 2005), the empirical validity of the adjusted- actuarial approach is far from established.

Indeed, the development of the dynamic risk instrument SONAR in Hanson and Harris’ (2000) research contained limita- tions such as invalid items, while its reﬁnement in a later study (Hanson, Harris, Scott, & Helmus, 2007) had little eﬀect on its capacity to add incremental validity to actuarial predictions. While the method holds some promise for evaluating changes to an oﬀender’s risk, further and better research is needed to justify the use of this method in a legal context.

Risk assessment tools As noted, a risk assessment should be based upon the best available methodology (Mercado & Ogloﬀ, 2007). Clearly, meeting 548D. J. Doyle et al. this requirement necessitates the use of the best available risk assessment tools. Posi- tively, the results showed that the majority of clinicians used valid and reliable struc- tured tools, across actuarial and SPJ methodologies.

The Static-99 was utilized in almost all reports, which reﬂects its status as a well- validated tool with reliably moderate degrees of accuracy (Hanson & Morton- Bourgon, 2007). However, used only in half of the reports were SPJ tools, the RSVP and SVR-20. Although SPJ tools are relatively recent, they have been validated in a number of studies with promising results (Craig et al., 2004; de Vogel et al., 2004; Macpherson, 2003). Indeed, Hanson and Morton-Bourgon’s (2007) meta-analy- sis of the accuracy of risk assessment instruments revealed that the strongest single predictor of sexual recidivism was a measure of SPJ (i.e., de Vogel et al., 2004).

The SPJ approach, unlike actuarial tools, can assist the clinician in the formulation of the nature of the risk posed by the oﬀender; it has also been recommended for those wishing to understand their cases in depth (Hanson & Morton-Bourgon, 2007).

Last, a number of assessors utilized the dynamic risk tool SONAR (Hanson & Harris, 2000; Hanson & Harris, 2001). As discussed previously, this risk instrument, and the adjusted-actuarial method to which it belongs, lacks the suﬃcient em- pirical base to justify its use in preventive detention proceedings.

Communicating Static-99 results The utility of a risk assessment tool is realized only when it is correctly adminis- tered, accurately interpreted, and its results are eﬀectively communicated. Concer- ningly, the results showed that the utility of the Static-99 was too often undermined by inadequate reporting of outcome in- formation, erroneous reporting of its re- sults, and its incorrect interpretation.First, omitted from a signiﬁcant num- ber of reports were the probability percen- tages of recidivism associated with the oﬀender’s risk score and the samples’ recidivism base rate upon which the prob- ability percentages were determined. This information is imperative to understanding and contextualizing the tool’s risk rating; its omission disallows judicial decision makers the necessary information to fully understand the descriptive and relative nature of the Static-99’s rating in relation to future sex oﬀending risk.

Several reports also expressed the prob- ability estimates associated with an oﬀen- der’s Static-99 score as indicating the oﬀender’sspeciﬁcrisk of reoﬀending (i.e., ‘‘There is a 4 in 10 chance that [the oﬀender] will reoﬀend within 5 years’’). More reports still were unclear about this relationship between the probability estimate and the oﬀender’s speciﬁc reoﬀence risk (i.e., ‘‘[the oﬀender’s] risk is quantiﬁed as a 40% likelihood of reoﬀending over 5 years from a sample of similar oﬀenders in Canada and the United Kingdom’’). The probability estimate associated with the oﬀender’s score refers to the recidivism percentages of agroupof sexual oﬀenders.

To apply this group-based percentage to the individual is wrong. Given the gravity of the decision to be made by court partly, but necessarily, based upon the expert’s risk assessment, this incorrect interpretation of the Static-99 risk tool is a glaring error.

A number of mistakes were also made in the direct reporting of the probability estimates themselves. Contrary to the tool’s manual (Harris, Phenix, Hanson, & Thorn- ton, 2003), a handful of evaluators applied the uncollapsed recidivism percentages to oﬀenders whose risk score was higher than 6. In so doing, the oﬀender’s risk for reoﬀending has been erroneously inﬂated.

In other reports, recidivism percentages were incorrectly quoted. For example, a number of reports stated the 15-year recidivism estimate for the high-risk An Analysis of Australian Forensic Reports549 category as ‘‘54%’’ when the instrument’s manuals and publications note this as ‘‘52%’’ (Hanson & Thornton, 1999). In some reports, the probability estimates were rounded up (i.e., from 39% to 40%). While these may be relatively minor errors, given the role that the exact numerical probability estimate may play in the judiciary’s decision on whether the oﬀender meets the threshold level of risk to warrant an order (see RJE v Secretary to the Department of Justice, 2008; TSL v Secretary to the Department of Justice, 2007), it is a fundamental requirement that the correct percentages are communicated to the courts.

Finally, users of the Static-99 consid- ered the eﬀect of the oﬀenders’ age on the validity of the actuarial assessment on very few occasions. This is contrary to the research evidence. The literature indicates that actuarial instruments insuf- ﬁciently capture the decline in recidivism risk associated with advanced age (Ogloﬀ & Doyle, 2009). Indeed, adjusted age- related probability estimates have been available for some years (Hanson, 2005), thus clincians have available to them empirically-supported guidelines to con- sider the impact of oﬀenders’ age on their risk.

It has been argued that the incorrect use of a recognized risk measure is potentially worse than not using a measure at all for ‘‘erroneous and misleading conclusions may be drawn that appear to have the weight of scientiﬁc research behind them and therefore carry an undeserved weight in legal proceedings’’ (Vess, 2009b, p.186).

The standard of practice in the use of the Static-99 must be raised.

Stating the limitations to risk assessment Limitations associated with risk assessment tools or the risk assessment enterprise more generally, were infrequently stated. There a number of pertinent limitations to theassessment of risk for future sexual vio- lence, as outlined here and elsewhere (Ogl- oﬀ & Doyle, 2009). Failure to convey the limits to the technology upon which one’s expert opinion rests, contravenes the pro- fessional’s ethical obligations and invites the potential for the court to accord undeserved weight to the risk outcome than is warranted (Glazebrook, 2010).

Risk factors A number of reports identiﬁed factors outside of a formal risk assessment proce- dure the author believed to be associated with an increased risk for sexual recidivism.

A majority of such factors were static and dynamic factors themed around the oﬀen- der’s sexual criminal history, sexual de- viance, and maladaptive interpersonal functioning and social supports. These factors have consistently been identiﬁed as empirically associated with sexual reci- divism (Hanson & Bussie` re, 1998; Hanson & Morton-Bourgon, 2005), thus reﬂecting an emerging concordance of opinion on risk. However, conversely, a number of reports also included risk factors that were identiﬁed by evaluators as being associated with risk that have equivocal or no empirical support (Hanson & Bussie` re, 1998; Hanson & Morton-Bourgon, 2005).

For example, a number of reports identi- ﬁed factors such as denial, a lack of victim empathy, low treatment motivation, and history of violent oﬀending, to indicate elevated risk, when in fact such factors do not have a well-accepted empirically sup- ported relationship to the outcome being predicted (Hanson & Bussie` re, 1998).

Simply, one cannot justify the use of such risk factors.

Inter-Rater Reliability The inter-rater reliability of risk assessment scores on the Static-99 was good, though somewhat lower than previous research 550D. J. Doyle et al. (Bartosh, Garby, Lewis, & Gray, 2003; Harris, 2003; Levenson, 2004). This indi- cates that clinicians will consistently pro- vide comparable assessments of the likelihood of sexual reoﬀence risk based on this instrument. The inter-rater relia- bility for Axis I paraphilic diagnoses was high; thus clinicians are reliably identifying the important psychopathological con- structs that are linked to reoﬀence risk.

Finally, the level of agreement between the evaluators regarding the ﬁnal risk rating was good. This result is signiﬁcant for it indicates that there is a consensus in relevant areas relating to risk assessment.

Communication of Final Risk Rating Almost all reports communicated a ﬁnal opinion on risk using the categorical method (i.e., high, medium, and low). A uniform approach to risk communication is clearly helpful for the task the court is engaged in. Last, some reports included additional descriptors such as ‘‘some risk,’’ ‘‘virtually certain risk,’’ ‘‘unacceptably high,’’ or ‘‘very high’’ in addition to their conclusion of high risk. These additional statements are ambiguous, potentially mis- leading, and likely to contribute an un- necessary element of confusion to considerations of risk.

Recommendations for Best Practice The ﬁndings of this investigation suggest there is substantial room for improvement in how clinicians assess risk for sexual violence and communicate their ﬁndings.

In order to strengthen the reliability and validity of expert opinion in this area and preserve professional standards, a number of recommendations can be put forward.

(1) The use of an unstructured clinical judgement approach to the assess- ment of risk is invalid, and there- fore has no probative value andshould not be relied upon exclu- sively in these assessments.

(2) Clinicians are cautioned against adjusting actuarially derived risk ratings based on dynamic risk vari- ables until more and better research provides an empirically defensible reason for doing so. Dynamic risk factors are useful in understanding oﬀending in an individual but as yet have limited validity in predicting risk over the long-term.

(3) The actuarial and SPJ methods are valid and complementary ap- proaches to risk assessment. Ac- tuarial assessments are wisely used to anchor the risk assessment, given the empirically robust relationship between static risk factors and future sexual violence. The SPJ method complements this approach by incorporating dynamic and idio- graphic risk information into a comprehensive evaluation of the possible nature of future sexual violence and provides targets for risk management. It would be appropriate for clinicians to utilize both when assessing risk for future sexual violence in applied assess- ments such as preventive detention proceedings.

(4) The types of risk assessment tools recommended for use are simply those that have the greatest evi- dence base. Accordingly, the Static- 99 remains the most reliable and best validated actuarial measure.

The RSVP and its predecessor the SVR-20 are very promising SPJ tools and will enable the clinician to understand their cases in much greater detail. The PCL-R is also a valid measure of the construct of psychopathy, which, given its rela- tionship to SVR, requires evalua- tion in an assessment of risk for future sexual violence. An Analysis of Australian Forensic Reports551 (5) The results of a properly conducted risk assessment must be eﬀectively communicated. When reporting the results of the Static-99 clinicians are advised to qualify comparative ca- tegorical labels such as high risk with the associated probability esti- mates of recidivism. Additionally, clinicians must report the correct probability percentages, consider the eﬀect of the oﬀender’s age on the validity of the actuarial result, and be very careful to not assign the probability estimate of recidivism to the speciﬁc oﬀender.

(6) The base rate of recidivism associated with an actuarial instrument’s test sample must also be communicated.

This is because the evidentiary value of the oﬀender’s probability of reci- divism is dependent on the base rate.

(7) Clinicians must know and convey the limitations to the state of knowledge in the ﬁeld of risk assessment. Failure to do so vio- lates ethical obligations and poten- tially gives the court the wrong impression about the predictive ability of the available technology.

(8) When communicating ﬁnal opinions on risk, evaluators should employ the categorical method (e.g., high, med- ium, low). Deﬁned conventionally, medium/moderate risk would be equal to the base rate of recidivism associated with the individual oﬀen- der, high risk is signiﬁcantly higher than the base rate, and low risk is signiﬁcantly lower than the base rate. Evaluators are also cau- tioned from describing an oﬀender’s risk as ‘‘unacceptable,’’ ‘‘signiﬁcant,’’ or ‘‘likely.’’ Such terms invite un- necessary ambiguity to the process of determining an oﬀender’s risk potential.

(9) Forensic clinicians need to be re- minded that research has identiﬁednumerous factors to be empirically associated with sexual reoﬀence risk. To identify factors that lack such empirical support is unjustiﬁ- able. Clinicians are recommended to remain well versed in the con- temporary research literature on risk assessment. Keeping up-to- date with scientiﬁc advances and debates within the ﬁeld will protect the clinicians’ opinion on risk from serious criticism and a scathing cross-examination.

Limitations and Future Directions This present study has limitations that should be considered when interpreting these ﬁndings. First, the sample size was smaller than expected following Queens- land’s decision not to participate in the study. Therefore, the results apply only to those clinicians preparing reports under this legislation in New South Wales, Western Australia, and Victoria. Second, the dis- ciplines of psychiatry and psychology were not evenly represented and some evaluators authored more reports in the sample than others. Accordingly, some of the results may be more relevant to a particular discipline, or author. Nevertheless, the sample of reports is a valid representation of reports tendered in these matters across the participating jurisdiction. Another lim- itation was the small number of cases available for the inter-rater reliability analyses.

Additional research is required in this area. An extension of this analysis to Queensland is warranted to provide a truly national assessment of the standard of practice of risk assessment in this legal context. Further, expanding research into the areas of clinicians’ decision-making processes in evaluating risk and how out- comes derived from multiple risk assess- ment tools and methods are integrated into a ﬁnal opinion on risk, is needed. 552D. J. Doyle et al. Conclusion This investigation described the risk assess- ment practices of forensic clinicians follow- ing an analysis of sex oﬀender assessment reports submitted under Australia’s pre- ventive detention legislation. As an analysis of clinicians’ actual practices, via their reports, this research represents a metho- dological advancement of previous investi- gations into the clinical practices of mental health professionals.

The results of this investigation were mixed. On the one hand, the ﬁndings indicated that there is a consensus in relevant areas relating to risk assessment resulting in regular agreement between clinicians with regard to the ﬁnal risk assessment outcome. Furthermore, the consistent use of structured risk tools is an encouraging sign that the hard science of the ﬁeld is translating into applied practice. However, beyond these relatively positive ﬁndings, more concerning results were found. Too many clinicians used unreliable methods of risk assessment, erroneously reported the results of a risk instrument, and failed to eﬀectively com- municate risk assessment outcomes. Too few clinicians stated the limitations that pertain to the science of risk assessment. In short, the standard of the practice of risk assessment for future sexual oﬀending must be raised. Clinical modesty and professional rigour are required.

The implications of these ﬁndings extend beyond recommendations for im- proving clinical practice. Preventive deten- tion proceedings involve fundamental questions of human rights and community safety. Under this legislation courts are faced with the unenviable task of balancing the human rights of oﬀenders with the risk to community safety posed by such oﬀen- ders. A less than competent risk assessment and report unnecessarily complicates this task with potentially deleterious conse- quences for the public and the oﬀender.It is hoped that this research will be used by clinicians, judges, and the legal fraternity to raise the level of practice in this area, so that the court can be assisted in its eﬀorts to achieve that balance between the civil liberties of oﬀenders and the right of the public to be kept safe from undue risk.

Note 1. However, two psychologists were approach- ing completion of their doctoral degrees in psychology.

References Allan, A., Martin, M., & Allan, M. (2000).

Assessments for the courts: A survey of Australian psychologists.Psychiatry, Psy- chology, and Law,7, 150–158.

Allnutt, S.H., & Chaplow, D. (2000). General principles of forensic report writing.Aus- tralian and New Zealand Journal of Psy- chiatry,34, 980–987.

Amenta, A.E. (2005).The assessment of sex oﬀenders for civil commitment proceedings:

An analysis of report content. Unpublished doctoral dissertation, University of Mis- souri, Columbia.

American Psychiatric Association. (2000).Di- agnostic and Statistical Manual of Mental Disorders(4th ed.). Washington, DC:

American Psychiatric Association.

Australian Psychological Society. (2007).Code of Ethics. Melbourne: Australian Psycholo- gical Society.

Barbaree, H.E., Seto, M.C., Langton, C.M., & Peacock, E.J. (2001). Evaluating the pre- dictive accuracy of six risk assessment instruments for adult sex oﬀenders.Crim- inal Justice and Behavior,28, 490–521.

Bartko, J.J. (1966). The intraclass correlation coeﬃcient as a measure of reliability.

Psychological Reports,19, 3–11.

Bartosh, D.L., Garby, T., Lewis, D., & Gray, S.

(2003). Diﬀerences in the predictive validity of actuarial risk assessments in relation to sex oﬀender type.International Journal of Therapy and Comparative Criminology,47, 422–438.

Berlin, F.S., Galbreath, N.W., Geary, B., & McGlone, G. (2003). The use of actuarials at civil commitment hearings to predict the likelihood of future sexual violence.Sexual Abuse: A Journal of Research and Treat- ment,15, 377–382. An Analysis of Australian Forensic Reports553 Bloom, M., Fischer, J., & Orme, J.G. (1999).

Evaluating practice: Guidelines for the ac- countable professional(3rd ed.). Boston:

Allyn and Bacon.

Boer, D.P., Hart, S.D., Kropp, P.R., & Web- ster, C.D. (1997).Manual for the Sexual Violence Risk-20. Vancouver, BC: The British Columbia Institute Against Family Violence.

Cohen, J. (1988).Statistical power analysis for the behavioural sciences(2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.

Craig, L.A., Browne, K.D., & Stringer, I.

(2004). Comparing sex oﬀender risk assess- ment measures on a UK sample.Interna- tional Journal of Oﬀender Therapy and Comparative Criminology,48, 7–27.

Davis, M.R., & Ogloﬀ, J.R.P. (2008). Key considerations and problems in assessing risk for violence. In D.C.R. Zukauskiene (Ed.),Psychology and law: Bridging the gap(pp. 191–210). Aldershot, UK:

Ashgate.

de Vogel, V., de Ruiter, C., van Beek, D., & Mead, G. (2004). Predictive validity of the SVR-20 and Static-99 in a Dutch sample of treated sex oﬀenders.Law and Human Behavior,28, 235–251.

Douglas, K.S., & Skeem, J.L. (2005). Violence risk assessment. Getting speciﬁc about being dynamic.Psychology, Public Policy, and Law,11, 347–383.

Doyle, D.J., & Ogloﬀ, J.R.P. (2009). Calling the tune without the music: A psycho-legal analysis of Australia’s post-sentence legisla- tion.Australian and New Zealand Journal of Criminology,42, 179–203.

Doyle, D.J., Ogloﬀ, J.R.P., & Thomas, S.D.M.

(in press). Designated as dangerous: Char- acteristics of sex oﬀenders subject to post- sentence orders in Australia.Australian Psychologist.

Freckelton, I.R., & Selby, H.M. (2009).Expert evidence: Law, practice, procedure and policy (4th ed.). Sydney: Thomson-Reuters.

Glazebrook, S. (2010). Risky business: Predict- ing recidivism.Psychiatry, Psychology, and Law,17, 88–120.

Hanson, R.K. (2005).The validity of Static-99 with older sexual oﬀenders. Ottawa:

Public Safety and Emergency Preparedness Canada.

Hanson, R.K., & Bussie` re, M.T. (1998). Pre- dicting relapse: A meta-analysis of sexual oﬀender recidivism studies.Journal of Consulting and Clinical Psychology,66, 348–362.Hanson, R.K., & Harris, A. (2000).The Sex Oﬀender Need Assessment Rating (SO- NAR): A method for measuring change in risk levels. Ottawa: Public Safety and Emergency Preparedness Canada.

Hanson, R.K., & Harris, A.J.R. (2001). A structured approach to evaluating change among sexual oﬀenders.Sexual Abuse: A Journal of Research and Treatment,13, 105– 122. Hanson, R.K., Harris, A.J.R., Scott, T.-L., & Helmus, L. (2007).Assessing the risk of sexual oﬀenders on community supervision:

The dynamic supervision project. Ottawa:

Public Safety Canada.

Hanson, R.K., & Morton-Bourgon, K.E.

(2004).Predictors of sexual recidivism: An updated meta-analysis (User Report 2004– 02). Ottawa: Public Safety and Emergency Preparedness Canada, Retrieved Septem- ber 24, 2008 from http://ww2.pssp.gc.ca/ publications/corrections/pdf/200402_e.pdf.

Hanson, R.K., & Morton-Bourgon, K.E.

(2005). The characteristics of persistent sexual oﬀenders: A meta-analysis of recidi- vism studies.Journal of Consulting and Clinical Psychology,73, 1154–1163.

Hanson, R.K., & Morton-Bourgon, K.E. (2007).

The accuracy of recidivism risk assessments for sexual oﬀenders: A meta-analysis (User Report 2007–01). Ottawa: Public Safety and Emer- gency Preparedness Canada.

Hanson, R.K., Morton, K.E., & Harris, A.J.R.

(2003). Sexual oﬀender recidivism risk:

What we know and what we need to know.Annals of the New York Academy of Sciences,989, 154–166.

Hanson, R.K., & Thornton, D. (1999).Static-99:

Improving actuarial risk assessments for sex oﬀenders (User Report 1999–02).Ottawa:

Department of the Solicitor General of Canada.

Harris, A., Phenix, A., Hanson, R.K., & Thornton, D. (2003).Static-99 coding rules revised.On- tario: Solicitor General Canada.

Harris, G.T. (2003). Men in his category have a 50% likelihood, but which half is he in?

Comments on Berlin, Galbreath, Geary, and McGlone.Sexual Abuse: A Journal of Research and Treatment,15, 389–392.

Harris, G.T., Rice, M.E., & Quinsey, V.L.

(2008). Shall evidence-based risk assessment be abandoned?British Journal of Psychia- try,192, 154.

Hart, S.D., Kropp, P.R., & Laws, D.R. (2003).

The risk for sexual violence protocol (RSVP): Structured professional guidelines 554D. J. Doyle et al. for assessing risk of sexual violence. British Columbia, Vancouver: Mental Health, Law, and Policy Institute, Simon Fraser Univer- sity and British Columbia Institute on Family Violence.

Hart, S.D., Michie, C., & Cooke, D.J. (2007).

Precision of actuarial risk assessment in- struments.British Journal of Psychiatry, 199, s60–s65.

Heilbrun, K., & Collins, S. (1995). Evaluations of trial competency and mental state at time of oﬀense: Report characteristics.Profes- sional Psychology: Research and Practice, 26(61–67).

Heilbrun, K., O’Neill, M.L., Strohman, L.K., Bowman, Q., & Philipson, J. (2000).

Expert approaches to communicating vio- lence risk.Law and Human Behaviour,24, 137–148.

Keyzer, P., Pereira, C., & Southwood, S. (2004).

Pre-emptive imprisonment for dangerous- ness in Queensland under the Dangerous Prisoners (Sexual Oﬀenders) Act 2003: The constitutional issues.Psychiatry, Psychol- ogy, and Law,11, 244–253.

La˚ ngstro ¨m, N. (2004). Accuracy of actuarial procedures for assessment of sexual oﬀender recidivism risk may vary across ethnicity.

Sexual Abuse: A Journal of Research and Treatment,16, 107–120.

Larkin, E.P., & Collins, P.J. (1989). Fitness to plead and psychiatric reports.Medicine, Science, and Law,29, 26–32.

Levenson, J.S. (2004). Reliability of sexually violent pedator civil commitment criteria in Florida.Law and Human Behaviour,28, 357–368.

Macpherson, G. (2003). Predicting escalation in sexually violent recidivism: Use of the SVR- 20 and PCL: SV to predict outcome with non-contact recidivists and contact recidi- vists.Journal of Forensic Psychiatry and Psychology,14, 615–627.

Martin, M., Allan, A., & Allan, M.M. (2001).

The use of psychological test by Australian psychologists who do assessments for the courts.Australian Journal of Psychology,53, 77–82.

McSherry, B. (2005). Indeﬁnite and preventive detention legislation: From caution to an open door.Criminal Law Journal,29, 94– 110.

McSherry, B., and Keyzer, P. (2009).Sex oﬀenders and preventive detention.Sydney:

The Federation Press.

McSherry, B., Keyzer, P., & Freiberg, A. (2006).

Preventive detention for ‘dangerous‘ oﬀendersin Australia: A critical analysis and proposals for policy development: Report to the Criminology Research Council. Retrieved 19 October 2009, from http://www.crimino- logyresearchcouncil.gov.au/reports/200405- 03.pdf.

Mercado, C.C., Elbogen, E.B., Scalora, M., & Tomkins, A. (2001). Judgments of danger- ousness: Are sex oﬀenders assessed diﬀer- ently than civil psychiatric patients?

Psychiatry, Psychology, and law, 8, 146– 153.

Mercado, C.C., & Ogloﬀ, J.R. (2007). Risk and the preventive detention of sex oﬀenders in Australia and the United States.Interna- tional Journal of Law and Psychiatry,30, 49–59.

Mullen, P.E., & Ogloﬀ, J.R.P. (Eds.). (2009).

Assessing and managing the risk of violence towards others(Vol. 2). Oxford, UK:

Oxford University Press.

Ogloﬀ, J., & Daﬀern, M. (2004).The dynamic appraisal of situational aggression (DASA).

Melbourne: Centre for Forensic Behaviour- al Science, Monash University.

Ogloﬀ, J.R.P., & Davis, M.R. (2005). Assessing risk for violence in the Australian context.

In D. Chappell & P. Wilson (Eds.),Issues in Australian crime and criminal justice(pp.

301–338). Chatswood, NSW: LexisNexis, Butterworths.

Ogloﬀ, J.R.P., & Doyle, D.J. (2009). A clarion call: Caution and humility must be the theme when assessing risk for sexual vio- lence under post-sentence laws.Sexual Abuse in Australia and New Zealand,2, 59–69.

Petrella, R.C., & Poythress, N.G. (1983). The quality of forensic evaluations: An inter- disciplinary study.Journal of Consulting and Clinical Psychology,51, 76–85.

Quinsey, V.L., Harris, G.T., Rice, M.E., & Cormier, C.A. (2006).Violent oﬀenders:

Appraising and managing risk(2nd ed.).

Washington, DC: American Psychological Association.

RJE v Secretary to the Department of Justice [2008] VSCA 265.

Royal Australian and New Zealand College of Psychiatrists. (1998).Code of ethics(2nd ed.). Melbourne: RANZCP.

Scott, R. (2008). Case commentary: Risk assessment and sentencing of serious sex oﬀenders.Psychiatry, Psychology, and Law, 15, 188–201.

Sentencing Advisory Council. (2006).High-risk oﬀenders: Continued detention and An Analysis of Australian Forensic Reports555 supervision options. Melbourne, Victoria:

Sentencing Advisory Council.

Shrout, P.E., & Fleiss, J.L. (1979). Intraclass correlations: Uses in assessing rater relia- bility.Psychological Bulletin,86, 420–428.

Skeem, J., Golding, S., Cohn, N., & Berge, G.

(1998). The logic and reliability of expert opinion on competence to stand trial.Law and Human Behaviour,22, 519–547.

Smallbone, S., & Ransley, J. (2005). Legal and psychological controversies in the preven- tive incapacitation of sexual oﬀenders.

University of New South Wales Law Journal, 28, 299–305.

TSL v Secretary to the Department of Justice [2006] VSCA 199.

Vess, J. (2009a). Fear and loathing in public policy: Ethical issues in laws for sex oﬀenders.Aggression and Violent Behaviour, 14, 264–272.Vess, J. (2009b). Risk assessment of sexual oﬀenders for extended supervision orders in New Zealand: Basic principles and current practice.Journal of Child Sexual Abuse,18, 174–189.

Victoria, Legislative Assembly. (2005).

Parliament of Victoria Hansard, 24 February 2005. Retrieved 21 March 2010, from http:// tex.parliament.vic.gov.au/bin/texhtml?form¼ VicHansard.adv Wollert, R. (2006). Low base rates limit expert certainty when current actuarials are used to identify sexually violent predators: An application of Baye’s theorem.Psychology, Public Policy and Law,12, 56–85.

Wood, M., & Ogloﬀ, J. R. (2006). Victoria’s Serious Sex Oﬀenders Monitoring Act 2005:

Implications for the accuracy of sex oﬀender risk assessment.Psychiatry, Psychology and Law,13, 182–198. 556D. J. Doyle et al. Copyright of Psychiatry, Psychology & Law is the property of Routledge and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.

However, users may print, download, or email articles for individual use.