Predicting Sex Offender Recidivism. I. Correcting for Item Overselection and Accuracy Overestimation in Scale Development. II. Sampling Error-Induced Attenuation of Predictive Validity Over Base Rate Information

The authors demonstrate a statistical bootstrapping method for obtaining unbiased item selection and predictive validity estimates from a scale development sample, using data (N = 256) of Epperson et al. [2003 Minnesota Sex Offender Screening Tool-Revised (MnSOST-R) technical paper: Development, validation, and recommended risk level cut scores. Retrieved November 18, 2006 from Iowa State University Department of Psychology web site: http://www.psychology.iastate.edu/ approximately dle/mnsost_download.htm] from which the Minnesota Sex Offender Screening Tool-Revised (MnSOST-R) was developed. Validity (area under receiver operating characteristic curve) reported by Epperson et al. was .77 with 16 items selected. The present analysis yielded an asymptotically unbiased estimator AUC = .58. The present article also focused on the degree to which sampling error renders estimated cutting scores (appropriate to local [varying] recidivism base rates) nonoptimal, so that the long-run performance (measured by correct fraction, the total proportion of correct classifications) of these estimated cutting scores is poor, when they are applied to their parent populations (having assumed values for AUC and recidivism rate). This was investigated by Monte Carlo simulation over a range of AUC and recidivism rate values. Results indicate that, except for the AUC values higher than have ever been cross-validated, in combination with recidivism base rates severalfold higher than the literature average [Hanson and Morton-Bourgon, 2004, Predictors of sexual recidivism: An updated meta-analysis. (User report 2004-02.). Ottawa: Public Safety and Emergency Preparedness Canada], the user of an instrument similar in performance to the MnSOST-R cannot expect to achieve correct fraction performance notably in excess of what is achievable from knowing the population recidivism rate alone. The authors discuss the legal implications of their findings for procedural and substantive due process in relation to state sexually violent person commitment statutes and the Supreme Court's Kansas v. Hendricks decision regarding the constitutionality of such statutes.     

Differences in the predictive validity of actuarial risk assessments in relation to sex offender type

This study seeks to expand on the previously reported validity of the Static-99, RRASOR, MnSOST-R, and SORAG in predicting sexual recidivism utilizing a regional sample of offenders. The predictive validity of each test was determined utilizing subgroups of the sample based on each offender's known offense history. The effectiveness of each instrument varied depending on offender type. The Static-99 and SORAG were both significantly predictive of sexual, violent, and any recidivism for extra-familial child molesters, and all four tests were predictive of violent or any recidivism in this subgroup. For incest offenders, all four tests were at least moderately predictive of sexual recidivism, whereas the Static-99 and the SORAG were highly predictive of violent or any recidivism. None of the four tests established consistent predictive validity across recidivism categories in regard to rapists or hands-off offenders, however, the Static-99 and the SORAG were significant in terms of sexual recidivism.     

Improving Nonmetric Sex Classification for Hispanic Individuals

Current techniques used by forensic anthropologists for the identification of unknown human skeletal remains have largely been created using U.S. Black and White samples. When applied to Hispanics, these techniques perform poorly and can lead to misclassifications; consequently, there is an imperative need for population‐specific standards for Hispanics. This research examines the classification accuracies obtained by the original Walker (Am J Phys Anthropol, 136, 2008) and Klales et al. (Am J Phys Anthropol, 149, 2012) methods for nonmetric sex estimation and provides recalibrated regression equations specifically for Hispanics. Ordinal data were collected for five skull and three pelvic traits from a sample of 54 modern Hispanic individuals. Recalibration of the Klales et al. equation improved accuracy (90.3% vs. 94.1%), while recalibration of the Walker method equation decreased accuracy (81.5% vs. 74.1%), but greatly improved sex bias (22.2% vs. ?7.4%), thereby making the recalibrated equations more appropriate for use with Hispanics.     

Problems in expert deception detection and the risk of false confessions: no proof to the contrary in Levine et al. (2014)

Lie-detection research has shown that observers who rely on nonverbal cues or on verbal cues correctly classify on average 54% of truth tellers and liars. In addition, over the years, countless numbers of innocent people have made false confessions and, in analysing the problem, researchers have implicated both a suspect's vulnerability and the persuasive influence of certain police interrogation tactics. Levine et al. (2014) aim to contribute to these vast bodies of literature by reporting two studies purportedly showing that expert interviewers – when they are permitted to question interviewees – can achieve almost perfect accuracy without eliciting false confessions. We argue that theoretical and methodological aspects of these studies undermine the reliability and validity of the data reported, that as a result the studies do not contribute to the scientific literatures on lie detection and false confessions in any meaningful way, and that the results are dangerously misleading.     

Psychopathy, treatment behavior, and recidivism: an extended follow-up of Seto and Barbaree

Seto and Barbaree reported the unexpected finding that adult male sex offenders who scored higher on psychopathy and exhibited better behavior in treatment were almost four times more likely to commit a new serious offence than other offenders once released. The present study reexamined this sample after a longer follow-up time using more complete recidivism data from a national police database. Although psychopathy continued to be a significant predictor of general and serious recidivism, treatment behavior was no longer related to either general or serious recidivism, and there was no statistically significant interaction between psychopathy and treatment behavior. Additional analyses ruled out the possibility that the differences between studies could be accounted for by the different average length of follow-up. A direct comparison of the two sources of recidivism data showed that differences in recidivism between subgroups were reduced by using the more complete recidivism data.     

Effectiveness of sex offender treatment for psychopathic sexual offenders

Meta-analyses have suggested that sexual offender treatment (SOT) completion is associated with lowered sexual recidivism rates for convicted sexual offenders. The paucity of properly designed studies allows for the alternative explanation of less recidivism among treated samples as reflecting that lower risk offenders disproportionately self-select into treatment. A test of the "self-selection explanation" can occur by investigating treatment effect on known high-risk offenders. Psychopathy correlates with increased sexual recidivism risk, such that an exploration of the SOT effect on psychopathic offenders could clarify the accuracy of the self-selection hypothesis. Additionally, the debated degree to which psychopaths are treatable might obtain clarification by a research review. This article examines empirical findings concerning the effectiveness of SOT for psychopathic sexual offenders. Ten studies were found to meet the minimal quality standards used, stemming from only four data sources. Shortcomings of existing research precluded clear conclusions, though trends in the data are delineated.     

A Japanese computer-assisted facial identification system successfully identifies non-Japanese faces

The method developed by Yoshino et al. in [Forensic Sci. Int. 109 (2000) 225 and Jpn. J. Sci. Tech. Iden. 5 (2000) 9] and already being applied in Japan utilizes a three-dimensional (3D) physiognomic rangefinder combined with a computer-assisted superimposition system. Facial outlines can be compared between two-dimensional (2D) surveillance images and data extracted from 3D images obtained from the rangefinder. Also, the loci of potentially concordant features can be compared and differences measured. The method is largely objective and gives statistics for false positive/false negative findings. This recently developed method by Yoshino et al. is currently being introduced to the Japanese courts. To enable courts outside Japan to assess the admissibility of this new method, studies of non-Japanese faces have been undertaken and shown to produce similar low error rates. The present authors, therefore, consider the Yoshino method to be applicable in a non-Japanese context. As part of this study a comparison of morphological features between two ethnic groups has been undertaken using 3D measurements for the first time and will serve as the foundation for an anthropological database in the future.     

Exploring Agreement on Appropriate Responses to Domestic Violence and Sexual Trauma Across Victim Advocates,Mental Health Service Providers,and Substance Abuse Treatment Providers

Abstract

Research suggests that survivors of domestic violence and sexual assault are more at risk than the general population for substance abuse and/or mental health disorders (Kendler et al., 2000). Additionally, research suggests that survivors of both crimes are at risk to be “multiply diagnosed-have” a history of victimization, a mental health diagnosis, and abuse of alcohol and drugs (Steele & Rechberger, 2002). While advocates have called for a change in the treatment of trauma survivors to treat their multiple needs, the availability of collaborative care between victim advocates and behavioral health professionals remains questionable. Using survey data collected from victim advocates, mental health service providers, and substance abuse treatment providers, this study assesses the extent to which providers of different backgrounds agree on how to effectively and appropriately serve domestic violence and sexual assault survivors-the first step in achieving the goal of holistic care.     

The cognitive distortions of child molesters are in need of treatment

Abstract

This paper forms the second part of a debate led by Marshall, Marshall, and Kingston (2011) regarding the need to address so-called cognitive distortions in sexual offender treatment. In their paper, Marshall et al. argue that so-called cognitive distortions may not necessarily require intense and focused attention or challenge throughout treatment. In evaluating Marshall et al.'s arguments, we highlight some inherent differences in how both Marshall et al. and ourselves choose to define the term “cognitive distortion”. We surmise that these key definitional differences appear to account for many of the issues that we “debate”. In particular, for example, Marshall et al. focus their arguments regarding cognitive distortions more explicitly upon excuses, denial and minimisations, whereas we choose to focus upon schemas and higher-order belief structures. Thus, we argue that the broadness and vagueness of the term “cognitive distortion” can lend itself to quite different interpretations and research foci. We offer some alternative views to Marshall et al.'s position and advocate the consideration of cognitions with an aetiological role in offending. We conclude with some suggestions for future research and treatment.     

Some implications of non-reporting and self-reporting of crime on criminal justice research

Conclusions There are reasonable grounds for assuming that unreported crime is extensive. Inasmuch as questionnaires are the major source of data for self-reported crime and delinquency and Clark and Tifft have seriously challenged the validity of this data source, it seems these areas lack a valid and unequivocal basis for assessing theory. It seems rather pointless to pursue etiological studies until new, valid, and reliable sources of data are available for theory construction and testing. The contention that interdisciplinary research is needed in Criminology is strengthened by the Clark and Tifft study, which offers an instance of fruitful interdisciplinary collaboration. Other recognized examples of significant interdisciplinary effort include the works of Pasmanick,et.al.,²³ Srole,et. al.,²⁴ and Lindner,et. al. ²⁵ Students of crime and delinquency must break through the parochialism of unitary and doctrinaire commitment to their own disciplines. In particular, sociologists must learn increasingly to utilize and embrace, at least on the micro-level, the potentials offered by other sciences, especially the life sciences. An alternative to this proposal would be a largescale consecutive birth study; then, perhaps in a quarter of a century, we might have the data necessary to formulate tentative etiological conclusions.     

Validity of the mandibular canine index (MCI) in sex prediction: Reassessment in an Indian sample

The dentition's use in sex assessment has been explored and advocated owing to its strength and resistance to peri- and postmortem insults. Among the teeth, canines have shown the greatest sexual dimorphism; they are also known for greater resistance to pathological alterations and trauma. These factors have prompted some researchers to explore its use alone in sex prediction. Specifically the mandibular canine index (MCI), which is the ratio of the mesiodistal (MD) dimension of the lower right canine and the inter-canine arch width, was developed by Rao et al. [Forensic Sci. Int. 42 (1989) 249–254] who predicted sex correctly in ～86% of cases in a South Indian sample. These results were verified subsequently by other Indian researchers, both on southern and on northern Indians, although such success rates were not apparent in another South-Asian population as well as a European group. The present study re-examined the method in 203 Indians (103 males, 100 females) of age 19–32 years who originated from diverse states of the country, constituting all major religious and caste affiliations, and found relatively low sex estimation accuracy using the MCI (～51%). This is in spite of the MD dimension of canines as well as inter-canine arch width – the absolute measurements used for deriving the MCI – showing statistically significant sex differences (p < 0.05); also, the application of multivariate statistics, viz. logistic regression analysis, revealed higher accuracy levels of sex estimation for the absolute measurements (～62–66%) vis-à-vis the MCI (～50%). The results herein reinforce recent observations that ratios obtained from teeth, such as the MCI, do not reflect sexual dimorphism that may be present in absolute measurements. Furthermore, we believe that the basis for using MCI as a sex assessment tool is questionable, since it depends on the levels of sexual dimorphism between the absolute dimensions—low sexual dimorphism in one (e.g. MD measurement of canines) and proportionately high sexual differences in the other (e.g. inter-canine arch width), or vice versa, would lead to higher sex estimation accuracy of the MCI; on the other hand, relatively equal levels of sexual dimorphism in both would negate the ability of the MCI to accurately estimate sex. These suggest that the MCI has little utility in sex estimation and that its application should be restricted, if not discontinued altogether, in forensic and anthropological sex prediction.     

Niederhoffer's cynicism scale: Reliability and beyond

This article presents an empirical assessment of the reliability and predictive validity of a modified version of Niederhoffer's (1967) police officer cynicism scale. The version of the scale examined in the study has been the subject of considerable methodological evaluations (Regoli, 1976a; Regoli and Poole, 1979; Anson et al., 1980; Anson, 1983). The results of these studies have not been consistent; nor have they come to grips with the predictive validity of the scale.Forty police officers attending a regional metropolitan training academy completed the scale at six-week intervals. The first administration of the scale subsequently was correlated to police supervisory ratings six months after scoring the Niederhoffer instrument.Longitudinal data revealed that the scale was an unstable and internally fragmented measurement instrument. Simple correlations between two distinct administrations of the scale and evaluations of police officers by superior officers indicated that the instrument had little predictive validity.The directions of the correlations support a reference group theory of police officer cynicism. Officers scoring high on cynicism receive higher job evaluations by superiors in police organizations. These data provide empirical support for the argument that cynicism is functional in the police officer subculture and that it is positively reinforced by superiors and peers.     

A Validation Study of the Langley et al. (2017) Decision Tree Model for Sex Estimation

Langley et al. (2017) developed a sex estimation decision tree utilizing two traditional cranial traits (glabella and mastoid) and a new trait: zygomatic extension. This study aimed to test the reliability of their zygomatic extension scoring method and validate their sex estimation method. Ordinal score data were collected from 281 male and female U.S. White and Black individuals. The five traditional cranial traits were collected from physical specimens, while zygomatic extension was scored from 3D cranial models. Intra‐ and interobserver analyses carried out on a subsample of 30 individuals indicate good agreement between zygomatic scores. The decision tree correctly sexed 71.5% of the sample, but a strong sex bias (94.2% correct for females, 49.3% correct for males) severely limits the utility of this method. The Walker (2008) and Stevenson et al. (2009) methods produced higher accuracy rates (80.8% and 82.6%, respectively), although these methods also produced sex and ancestry biases.     

Pronasale Position: An Appraisal of Two Recently Proposed Methods for Predicting Nasal Projection in Facial Reconstruction

This study examines two recently proposed methods for predicting nose projection from the skull, those developed by Stephan et al. (Am J Phys Anthropol 122, 2003, 240) and Rynn et al. (Forensic Sci Med Pathol 6, 2010, 20). A sample of 86 lateral head cephalograms of adult subjects from Central Europe was measured, and the actual and predicted dimensions were compared. Regarding nose projection (the anterior and vertical position of the pronasale), in general, the method of Rynn et al. (Forensic Sci Med Pathol 6, 2010, 20) was found to perform better and with less error variance than the method of Stephan et al. (Am J Phys Anthropol 122, 2003, 240), but the mean difference between the actual and predicted values did not exceed 2.2 mm (6.5% of the actual dimension) in most of the variables tested. The vertical dimensions of the nose were predictable with greater accuracy than the horizontal dimensions. Although the mean error of both methods is not great in practice and thus both methods could be considered to be “accurate,” the real variance of error should not be overlooked.     

Multidimensional scaling and the analysis of sexual offence behaviour: a reply to Sturidsson et al.

Abstract

Sturidsson et al. (Psychology, Crime & Law, 12, 221–230, 2006) published an article on the replicability of early multidimensional scaling (MDS) behavioural research by Canter and Heritage (Journal of Forensic Psychiatry, 1, 185–212, 1990). Sturidsson et al. were unable to find a ‘readily apparent, simple interpretation’ of their MDS solution and therefore suggested that the MDS approach may be flawed as a device for interpreting sexual assault. The current paper explains how the methods employed by Sturidsson et al. represent an incorrect use of MDS, specifically in their input of Jaccard similarity data into the MDS (ALSCAL) procedure in SPSS as opposed to the correct MDS (PROXSCAL) procedure. A comparative example is used to illustrate how adopting the ALSCAL procedure will result in an inverse MDS solution, where objects appearing closer together are more dissimilar, making interpretation of the MDS plot meaningless in this context. The current paper discusses this methodological flaw in greater depth and in the context of behavioural investigative research using MDS in general. It concludes with some general guidelines in utilizing the MDS procedure in SPSS specific to criminal profiling research.     

A Developmental-Ecological Model of the Relation of Family Functioning to Patterns of Delinquency

Person-centered approaches to understanding delinquent and criminal careershave identified several distinct pathways or patterns of delinquent behavior(Gorman-Smith et al., 1998; LeBlanc and Kaspy, in press; Loeber et al.,1991; Loeber et al., 1993). In addition, research suggests that there maybe etiological variations that correspond to these different pathways(Gorman-Smith et al., 1998). That is, there may be different configurationsof risk factors associated with different types of delinquent and criminaloffending. If this is so, understanding these relations can have importantimplications for intervention and prevention. However, there have been fewstudies that examine how the configuration of risk factors may vary inrelation to different delinquency pathways. The current study brings aperson-centered analysis to examine how patterns of family functioningrelate to patterns of offending. In addition, this study contextualizesthese relations by examining how these relations vary as a function ofcommunity setting. This study expands upon previous research that hasidentified four basic patterns of delinquent behavior among a sample ofminority male adolescents living in poor urban neighborhoods (Gorman-Smithet al., 1998).     

Frontal sinuses for identification: quality of classifications, possible error and potential corrections

Many studies have examined the characteristics of the frontal sinuses and their use for forensic purposes, particularly when an individual is edentulous. One of the most widespread classification systems is that proposed by Yoshino et al. The aim of this study was to improve the performance of Yoshino's method for identifying unknown skeletal remains by replacing the first two morphological items, frontal sinus size and bilateral asymmetry, by SOR1 = left frontal sinus area/left orbit area, and SOR2 = right frontal sinus area/right orbit area. According to the bivariate distribution of SOR = (SOR1, SOR2) and available data, we also estimated the probability of positive misclassification.     

The antidepressant debate and the balanced placebo trial design: an ethical analysis

There is ongoing debate about whether randomized, placebo-controlled trials under a double-blind have reliably established the pharmacological efficacy of antidepressants. Numerous meta-analyses of antidepressant efficacy trials, e.g., Kirsch et al. [Kirsch, I., Moore, T. J., Scoboria, A., & Nicholls, S. (2002). The emperor's new drugs: An analysis of antidepressant medication data submitted to the U.S. food and drug administration. Prevention and Treatment, 5, Article 23. (Retrieved July 19, 2007 from http://journals.apa.org/prevention/volume5)], have shown a modest drug-placebo difference but methodological problems with standard trial design preclude a definitive conclusion that this difference results from specific biological effects of antidepressants or the nonspecific factors that have not been adequately excluded. Standard trial design assumes the additivity thesis of pharmacological efficacy, being the assumption that the specific or "true" magnitude of the pharmacological effect is limited to the difference between the drug and placebo responses in a standard trial. If the drug effects are as small as these meta-analyses suggest, then their clinical effectiveness is questionable. If the drug effects are actually larger but masked by placebo effects, then the additivity thesis is not valid and we risk false negative results with standard trial design. Kirsch et al. propose an alternative, four arm balanced placebo trial design (BPTD) that can accurately test the additivity thesis. The BPTD uses antidepressants, active placebos and the intentional deception of research subjects. My focal question is whether the BPTD is ethically defensible. I will explore two objections that can be raised against it: 1) lying to BPTD research subjects violates their autonomy and exploits their illness and 2) the BPTD may not enable us to test the additivity thesis with accuracy, i.e., it may contribute to the masking of drug effects that it aims to avoid. I argue that these objections support the conclusion that the BPTD is ethically indefensible.     

Individual Confidence Intervals Do Not Inform Decision-Makers About the Accuracy of Risk Assessment Evaluations

Some recent articles have proposed that the confidence interval for the predicted outcome of a single case can be used to describe the predictive accuracy of risk assessments (Hart et al. Br J Psychiat 190:60–65, 2007b; Cooke and Michie, Law Hum Behav 2009). Given that the confidence intervals for an individual prediction are very large, Cooke and colleagues have questioned the wisdom of applying recidivism rates estimated from group data to single cases. In this article, we argue that the confidence intervals for the recidivism outcome predicted for a single case will range between zero to one (i.e., be uninformative) when the outcome is dichotomous and the predicted probability is between .05 and .95. This is true by definition and limits the utility of using individual confidence intervals to measure predictive accuracy. Consequently, other quality indicators (many of which are non-quantitative) are needed to determine the accuracy and error of risk evaluations.