Evaluating the Predictive Validity of the Compas Risk and Needs Assessment System

By Tim Brennan, William Dieterich, B. Ehret. 2009

396 citations recorded by the source corpus.

Subfields: Clinical Psychology, Sociology and Political Science, Evaluation & Benchmarks, AI Ethics, Fairness & Society

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Clinical Psychology

In Sociology and Political Science

In Evaluation & Benchmarks

In AI Ethics, Fairness & Society

Citations in the corpus

Counterfactual Fairness (2017, background): “Machine learning has spread to fields as diverse as credit scoring [20], crime prediction [5], and loan assessment [25].”
Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead (2018, methodology): “Interestingly, if the COMPAS model were not proprietary, its documentation [29] indicates that it would actually be an interpretable predictive model.”
Please Stop Explaining Black Box Models for High Stakes Decisions (2018, methodology): “Consider the COMPAS proprietary recidivism risk prediction tool discussed above that is in widespread use in the U.S. Justice System for predicting the probability that someone will be arrested after their release [Brennan et al., 2009].”
A Causal Bayesian Networks Viewpoint on Fairness (2018, neutral): “The COMPAS general recidivism risk scale (GRRS) and violent recidivism risk scale (VRRS), the focus of ProPublica’s investigation, sought to leverage machine learning techniques to improve the predictive accuracy of recidivism compared to older RAIs such as the level of service inventory-revised [3]”
The age of secrecy and unfairness in recidivism prediction (2018, background): “There have been numerous debates about fairness in the literature, mainly stemming from a flawed analysis by the ProPublica group (1, 2) of data from Broward County FL, claiming that the proprietary prediction model COMPAS (Correctional Offender Management Profiling for Alternative Sanctions) (3) is”
Fairness in Decision-Making - The Causal Explanation Formula (2018, background): “…and statistics have been increasingly applied throughout a wide range of real-world decision-making scenarios, including in healthcare, law enforcement, education, and ﬁnance (Mahoney and Mohen 2007; Brennan, Dieterich, and Ehret 2009; Khandani, Kim, and Lo 2010; Sweeney 2013; Angwin et al. 2016).”
An algorithm for removing sensitive information: Application to race-independent recidivism prediction (2017, background): “…of individuals’ future behavior are used to inform judges regarding pre-trial release and bail setting, sentencing, and parole [Phillips et al., 2016, Dieterich et al., 2016, Brennan et al., 2009, Cunningham and Sorensen, 2006, Dvoskin and Heilbrun, 2001, Quinsey et al., 1998, Berk et al., 2009].”
Nuanced Metrics for Measuring Unintended Bias with Real Data for Text Classification (2019, methodology): “As machine learning models are being used for an ever-expanding set of use cases, many have raised concerns about the potential negative impact of unintended identity-based bias that could be present in the models [3, 8].”
A statistical framework for fair predictive algorithms (2016, background): “Examples include hiring [8], predictive policing [13], pre-trial risk assessment of recidivism[6, 2], and risk of violence while incarcerated [5].”
The predictive validity of risk assessment tools for female offenders: A systematic review (2015, methodology): “Of the studies included in the review, seven were conducted in the US, four were undertaken in Canada, three were undertaken in the Netherlands, and one study was conducted in the UK.”
Explaining models: an empirical study of how explanations impact fairness judgment (2019, methodology): “6% on processed data, comparable to reports on the accuracy of COMPAS system [6, 21].”
Who is afraid of black box algorithms? On the epistemological and ethical basis of trust in medical AI (2021, background): “57 58 degree of surveyability is, indeed, possible. ii This contrast can be illustrated with a simple example.”
Fair Kernel Learning (2017, background): “New social and economic activities massively exploit big data and machine learning algorithms to do inferences, and they decide on the best curriculum to fill in a position [15], to determine wages and in pre-trial risk assessment [4, 9], and to evaluate risk of violence [8].”
Attenuating Bias in Word Vectors (2019, background): “Word vector embeddings as used in machine learning towards applications which signiﬁcantly aﬀect people’s lives, such as to assess credit [11], predict crime [5], and other emerging domains such judging loan applications and resumes for jobs or college applications.”