A Multiple Models Approach to Assessing Recidivism Risk

  title={A Multiple Models Approach to Assessing Recidivism Risk},
  author={Eric Silver and Lynette Chow-Martin},
  journal={Criminal Justice and Behavior},
  pages={538 - 568}
This study used a large recidivism data set to develop and validate a multiple models tool for predicting recidivism risk. Consistent with prior research, the authors found that the multiple models tool was more accurate than tools built using the traditional single-model approach. In addition, they demonstrated that the predicted recidivism rates produced by the multiple models tool could be summarized in a usable format consisting of four to five statistically distinct risk classes offering… 

Figures and Tables from this paper

Interpretable Classification Models for Recidivism Prediction
A new method known as SLIM (Supersparse Linear Integer Models) is used to produce accurate, transparent, and interpretable models along the full ROC curve, which can be used for decision-making for many different use cases, since they are just as accurate as the most powerful black-box machine learning models.
Interpretable Classification Models for Recidivism Prediction May 26 , 2015
  • 2015
We investigate a long-debated question of how to create predictive recidivism models that are sufficiently accurate, transparent, and interpretable to use for decision-making. These models are used
Isolating modeling effects in offender risk assessment
ObjectivesRecent evolutions in actuarial research have revealed the potential increased utility of machine learning and data-mining strategies to develop statistical models such as
Optimizing predictive performance of criminal recidivism models using registration data with binary and survival outcomes
Results on the reconviction data from two sources suggest that both statistical and machine learning should be tried out for obtaining an optimal model, and a flexible black-box model can serve as a test whether important interactions are missing or other misspecification of the model are present and can thus provide more security in the modeling process.
Breaking the 'glass ceiling' of risk prediction in recidivism: an application of connectionist modelling to offender data
The present thesis explored the capability of connectionist models to break through the ‘glass ceiling’ of accuracy currently in operation in recidivism prediction (e.g., Yang, Wong, & Coid, 2010).
Using decision tree algorithms to screen individuals at risk of entry into sexual recidivism
•Decision-tree algorithms (DTA) can be used to examine the risk profiles of convicted offenders.•DTA highlight the importance of the offender's age at assessment as key risk factor.•DTA findings
Assessing Violence Risk in Stalking Cases: A Regression Tree Approach
A Classification and Regression Tree (CART) approach is applied to a sample of stalking offenders in order to help further the process of identifying and understanding risk assessment strategies.
A comparison of statistical models in predicting violence in psychotic illness.
Although classification trees can be suitable for routine clinical practice, because of the simplicity of their decision-making processes, their robustness and therefore clinical utility was problematic in this sample of patients with psychotic illness.
Prediction modelling for population conviction data
This monograph is concerned with different uses of prediction modelling on (Dutch) Judicial registration data. It covers the determining the usefulness, preservability of a prediction instrument used
Using Risk Assessment to Inform Sentencing Decisions for Nonviolent Offenders in Virginia
Virginia has abolished parole and adopted truth-in-sentencing guidelines for persons convicted of felonies. As part of this reform, the Virginia Criminal Sentencing Commission developed a method for


Constructing Actuarial Devices for Predicting Recidivism
A recent contribution to the prediction literature by Steadman et al. features a novel “iterative classification” procedure for constructing risk screening devices. In this article, the authors apply
A Classification Tree Approach to the Development of Actuarial Violence Risk Assessment Tools
This work proposes a classification tree rather than a main effects regression approach for actuarial violence risk assessment tools, and suggests that by employing two decision thresholds for identifying high- and low-risk cases, the use of actuarial tools to make dichotomous risk classification decisions may be further enhanced.
The consequences of error: Recidivism prediction and civil-libertarian ratios
Abstract The utility of recidivistic prediction is limited by the false positive problem: predictions of failure (recidivism) that do not occur. False negatives (predicted successes but observed
A Cautionary Note on the Use of Actuarial Risk Assessment Tools for Social Control
Actuarial risk assessment tools for predicting interpersonal violence and criminal recidivism have proliferated in recent years, promising to enhance their use in the day-to-day operations of the
Developing a clinically useful actuarial tool for assessing violence risk
A clinically useful actuarial method exists to assist in violence risk assessment and this work aims to increase the clinical utility of the ICT method by restricting the risk factors used to generate the actuarial tool to those commonly available in hospital records or capable of being routinely assessed in clinical practice.
The dimensions of dangerousness revisited
In this article, we extend our previous study on clinical predictions of violence using the Dangerous Behavior Rating Scale by increasing follow-up interval from 2 to 6 years and supplying new data
The effects of base rate and cutoff point choice on commonly used measures of association and accuracy in recidivism research
Several measures of association and accuracy commonly used in the recidivism literature are examined for their sensitivity to variations in the proportion observed to recidivate (the base rate) and
Selective incapacitation: Have we tried it? Does it work?
Abstract The selective incapacitation of individuals who pose a threat to society by their frequent criminal activity has been recently discussed widely by academicians, policy makers, and
Traditional views of incapacitation as a crime control strategy rely fundamentally on behavioral prediction. Most attempts at predicting offenders' behaviors have relied on simple dichotomous
Selective Incapacitation: Sentencing According to Risk
In October 1982, the Rand Corporation published Selective Incapacitation, a sentencing proposal based on seven years of research by a team of Rand researchers under the direction of Peter Greenwood.