Motivational Interviewing: Preparing People for Change

The history of risk assessment in criminal justice has been written on several occasions (Andrews & Bonta, 2003; Clements, 1996; Hollin, 2002). Here we assess progress since Andrews, Bonta, and Hoge’s (1990; Andrews, Zinger, et al., 1990) statement of the human service principles of risk-needresponsivity (RNR) and professional discretion. In those articles, the corrections-based terms of risk and need were transformed into principles addressing the major clinical issues of who receives treatment (higher risk cases), what intermediate targets are set (reduce criminogenic needs), and what treatment strategies are employed (match strategies to the learning styles and motivation of cases: the principles of general and specific responsivity). General responsivity asserts the general power of behavioral, social learning, and cognitive-behavioral strategies. Specific responsivity suggests matching of service with personality, motivation, and ability and with demographics such as age, gender, and ethnicity. Nonadherence is possible for stated reasons under the principle of professional discretion. Expanded sets of principles now include consideration of case strengths, setting of multiple criminogenic needs as targets, community-based, staff relationship and structuring skills, and a management focus on integrity through the selection, training, and clinical supervision of staff and organizational supports (Andrews, 2001). The review is conducted in the context of the advent of the fourth generation of offender assessment. Bonta (1996) earlier described three generations of risk assessment. The first generation (1G) consisted mainly of unstructured professional judgments of the probability of offending behavior. A

The Recent Past and Near Future of Risk and/or Need Assessment

The present meta-analysis integrates research from 1,435 studies on associations of parenting dimensions and styles with externalizing symptoms in children and adolescents. Parental warmth, behavioral control, autonomy granting, and an authoritative parenting style showed very small to small negative concurrent and longitudinal associations with externalizing problems. In contrast, harsh control, psychological control, authoritarian, permissive, and neglectful parenting were associated with higher levels of externalizing problems. The strongest associations were observed for harsh control and psychological control. Parental warmth, behavioral control, harsh control, psychological control, autonomy granting, authoritative, and permissive parenting predicted change in externalizing problems over time, with associations of externalizing problems with warmth, behavioral control, harsh control, psychological control, and authoritative parenting being bidirectional. Moderating effects of sampling, child's age, form of externalizing problems, rater of parenting and externalizing problems, quality of measures, and publication status were identified. Implications for future research and practice are discussed. (PsycINFO Database Record

Associations of parenting dimensions and styles with externalizing problems of children and adolescents: An updated meta-analysis.

The nascent field of fair machine learning aims to ensure that decisions guided by algorithms are equitable. Over the last several years, three formal definitions of fairness have gained prominence: (1) anti-classification, meaning that protected attributes---like race, gender, and their proxies---are not explicitly used to make decisions; (2) classification parity, meaning that common measures of predictive performance (e.g., false positive and false negative rates) are equal across groups defined by the protected attributes; and (3) calibration, meaning that conditional on risk estimates, outcomes are independent of protected attributes. Here we show that all three of these fairness definitions suffer from significant statistical limitations. Requiring anti-classification or classification parity can, perversely, harm the very groups they were designed to protect; and calibration, though generally desirable, provides little guarantee that decisions are equitable. In contrast to these formal fairness criteria, we argue that it is often preferable to treat similarly risky people similarly, based on the most statistically accurate estimates of risk that one can produce. Such a strategy, while not universally applicable, often aligns well with policy objectives; notably, this strategy will typically violate both anti-classification and classification parity. In practice, it requires significant effort to construct suitable risk estimates. One must carefully define and measure the targets of prediction to avoid retrenching biases in the data. But, importantly, one cannot generally address these difficulties by requiring that algorithms satisfy popular mathematical formalizations of fairness. By highlighting these challenges in the foundation of fair machine learning, we hope to help researchers and practitioners productively advance the area.

The Measure and Mismeasure of Fairness: A Critical Review of Fair Machine Learning.

The effects of correctional interventions on recidivism have important public safety implications when offenders are released from probation or prison. Hundreds of studies have been conducted on those effects, some investigating punitive approaches and some investigating rehabilitation treatments. Systematic reviews (meta-analyses) of those studies, while varying greatly in coverage and technique, display remarkable consistency in their overall findings. Supervision and sanctions, at best, show modest mean reductions in recidivism and, in some instances, have the opposite effect and increase reoffense rates. The mean recidivism effects found in studies of rehabilitation treatment, by comparison, are consistently positive and relatively large. There is, however, considerable variability in those effects associated with the type of treatment, how well it is implemented, and the nature of the offenders to whom it is applied. The specific sources of that variability have not been well explored, but some princ...

The Effectiveness of Correctional Rehabilitation: A Review of Systematic Reviews

Successful prosecution of state-level domestic violence cases typically requires participation from the victim. However, given the unique factors associated with these cases, the assistance of a victim advocate may be beneficial in maintaining victim cooperation. This research examines 353 cases to determine if victim cooperation and case disposition could be predicted by victim advocacy, victim injuries, defendant's use of a weapon, and the presence of witnesses. Findings from logistic regression analysis suggest that victim assistance was significantly related to both victim cooperation and case disposition. Further, victim cooperation and the presence of witnesses significantly increased the likelihood of a conviction.

Predictors of Domestic Violence Prosecution in a State Court

Recent trends in corrections have mandated the adoption and use of risk and need assessments for offenders. Research indicates that many correctional agencies around the country either currently use or are in the process of implementing risk and need assessment instruments. One example of an instrument that is being implemented on a wide scale is the Level of Service Inventory–Revised (LSI-R). Data from Multi-Health Systems, Inc. (MHS), the company that markets the LSI-R, indicate that more than 600 agencies in United States currently use this risk/need assessment tool. While increasing use of objective classification instruments is encouraging, simultaneously there are growing concerns regarding the effective implementation of these “third-generation” risk/need assessment tools. The authors of the LSI-R, in discussing risk assessment in general, express concern about the ability of correctional practitioners to understand and properly administer risk assessment instruments. Research by Bonta, Bogue, Crowley, and Motiuk (2001) that investigated the implementation of the LSI-R indicated error rates that were of concern. Their research also indicated that these error rates could be reduced through the use of training prior to implementation. While currently there is a lack of quality assurance research conducted on other instruments such as the Wisconsin Risk and Need scales or the Salient Factor Score, it is conceivable that errors occur in these instruments as well. This may be particularly true when considering the potential for agencies, which may be in a rush to implement best practices, to skip necessary, preliminary research and training.

Empirical Evidence on the Importance of Training and Experience in Using the Level of Service Inventory-Revised

https://www.jstor.org/stable/info/10.1525/fsr.2015.27.4.216

Pretrial Risk Assessment: Improving Public Safety and Fairness in Pretrial Decision Making

Alexander M. Holsinger Criminal Justice Coordinator, Johnson County, KS Christopher T. Lowenkamp Administrative Office of the U.S. Courts Edward J. Latessa University of Cincinnati Ralph Serin Carleton University, Ottawa Thomas H. Cohen Administrative Office of the U.S. Courts Charles R. Robinson Administrative Office of the U.S. Courts Anthony W. Flores California State University, Bakersfield Scott W. VanBenschoten Administrative Office of the U.S. Courts

A Rejoinder to Dressel and Farid: New Study Finds Computer Algorithm Is More Accurate Than Humans at Predicting Arrest and as Good as a Group of 20 Lay Experts 1

Jurisdictions at every level throughout the US are paying an increasing amount of attention to pretrial case processing The primary areas of attention appear to be on risk assessment development and classification, the effects of pretrial detention, and the effectiveness of various strategies that may impact a defendant’s failure to appear for their assigned court dates The current study is a randomized experimental trial designed to test the effects of court notification strategies, using failure to appear (FTA) as the primary outcome of interest Our findings do not reveal a palpable effect for court notification strategies (telephone calls, and text messaging, with other conditions layered in), but do indicate and reinforce the utility of an actuarial method of risk classification when predicting likelihood of FTA

Alexander M. Holsinger

Papers

Predictors of Domestic Violence Prosecution in a State Court

Empirical Evidence on the Importance of Training and Experience in Using the Level of Service Inventory-Revised

Pretrial Risk Assessment: Improving Public Safety and Fairness in Pretrial Decision Making

A Rejoinder to Dressel and Farid: New Study Finds Computer Algorithm Is More Accurate Than Humans at Predicting Arrest and as Good as a Group of 20 Lay Experts 1

Assessing the Effects of Court Date Notifications within Pretrial Case Processing