Fairness in Criminal Justice Risk Assessments: The State of the Art

By Richard A. Berk, Hoda Heidari, Shahin Jabbari, Michael Kearns, Aaron Roth. 2017

1,137 citations recorded by the source corpus.

Objectives: Discussions of fairness in criminal justice risk assessments typically lack conceptual precision. Rhetoric too often substitutes for careful analysis. In this article, we seek to clarify the trade-offs between different kinds of fairness and between fairness and accuracy. Methods: We draw on the existing literatures in criminology, computer science, and statistics to provide an integrated examination of fairness and accuracy in criminal justice risk assessments. We also provide an empirical illustration using data from arraignments. Results: We show that there are at least six kinds of fairness, some of which are incompatible with one another and with accuracy. Conclusions: Except in trivial cases, it is impossible to maximize accuracy and fairness at the same time and impossible simultaneously to satisfy all kinds of fairness. In practice, a major complication is different base rates across different legally protected groups. There is a need to consider challenging trade-offs. These lessons apply to applications well beyond criminology where assessments of risk can be used by decision makers. Examples include mortgage lending, employment, college admissions, child welfare, and medical diagnoses.

Subfields: Economics and Econometrics, Sociology and Political Science, Evaluation & Benchmarks, Safety, Robustness & Interpretability

How researchers see this paper

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

In Economics and Econometrics

In Sociology and Political Science

In Evaluation & Benchmarks

In Safety, Robustness & Interpretability

Citations in the corpus

Measurement and Fairness (2019, background): “explicitly consider how to minimize or compensate for historical injustices [11]. This nontrivial measurement task is deeply complex and requires signicant debate about justice, policy, and fairness [8, 13, 15, 37]. 4.2 Value-Added Modeling in Education We now look to the well-known example of value”
Counterfactual Fairness (2017, background): “These criteria can be incompatible in general, as discussed in [1, 7, 22].”
Prediction-Based Decisions and Fairness: A Catalogue of Choices, Assumptions, and Definitions (2018, background): “[8] also consider a definition based on several of the above elements of the confusion matrix.”
Fairwashing: the risk of rationalization (2019, methodology): “To explain the fairness of machine models, various metrics have been proposed in recent years (Narayanan, 2018; Berk et al., 2018).”
A Unified Approach to Quantifying Algorithmic Unfairness: Measuring Individual &Group Unfairness via Inequality Indices (2018, background): “5For making the pretrial release decisions, these two are the main criteria that the judges or the algorithms try to assess [6, 36].”
Local Rule-Based Explanations of Black Box Decision Systems (2018, methodology): “In fact, if the training data contains a number of biased decision records, or misleading classification examples due to data collection mistakes or artifacts, it is likely that the resulting algorithm inherits the biases and recommends discriminatory or simply wrong decisions 2 [6, 7].”
Interventions over Predictions: Reframing the Ethical Debate for Actuarial Risk Assessment (2017, background): “et al: identify six kinds of fairness, and show that not only these notions conflict with accuracy, but also with one another [10].”
Fair Regression: Quantitative Definitions and Reduction-based Algorithms (2019, background): “Copyright 2019 by the author(s). have been proposed (Berk et al., 2017; Chouldechova, 2017; Hardt et al., 2016; Kleinberg et al., 2017), leading to a variety of algorithms that aim to satisfy them (see, e.g., Corbett-Davies & Goel, 2018, for an overview of the literature).”
Efficient candidate screening under multiple tests and implications for fairness (2019, background): “Several simple impossibility results preclude simultaneously satisfying several combinations of these parities [4, 5, 13].”
Fairness Under Unawareness: Assessing Disparity When Protected Class Is Unobserved (2018, background): “Examples of such decisions include employment and compensation [14, 20], university admissions [4], and sentence and bail setting [3, 7, 16].”
50 Years of Test (Un)fairness: Lessons for Machine Learning (2018, background): “• Peterson and Novick’s “conditional probability and its con-verse” sufficiency [4], equalized odds [32], or conditional procedure accuracy [5], sometimes expressed as the conditional independence A ⊥ D | Y .”
An Empirical Study of Rich Subgroup Fairness for Machine Learning (2018, background): “, 2017] — see Berk et al. [2018] for more examples. These definitions are pervasive in large part because they are easy to check, although there are interesting computational challenges in learning subject to these constraints in the worst case — see e.g. Woodworth et al. [2017]. Unfortunately, thes”
A Causal Bayesian Networks Viewpoint on Fairness (2018, neutral): “Many such deﬁnitions have been found to be mathematically incompatible [7, 12, 14, 15, 29], and this has been viewed as representing an unavoidable trade-oﬀ establishing fundamental limits on fair machine learning, or as an indication that certain deﬁnitions do not map on to traditional social or…”
On Fairness and Calibration (2017, background): “Calibrated probability estimates are considered necessary for empirical risk analysis tools [4, 10, 12, 16].”
Avoiding Discrimination through Causal Reasoning (2017, background): “A review of various fairness criteria can be found in [8], where they are discussed in the context of criminal justice.”