Risk, Race, & Recidivism: Predictive Bias and Disparate Impact

By Jennifer Skeem, Christopher T. Lowenkamp. 2016

72 citations recorded by the source corpus.

Subfields: Clinical Psychology, Sociology and Political Science

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

Citations in the corpus

Fair Prediction with Disparate Impact: A Study of Bias in Recidivism Prediction Instruments (2016, methodology): “These standards have recently been applied to the COMPAS [3] and PCRA [4] instruments, with initial findings suggesting that there is evidence of predictive bias when it comes to gender, but not when it comes to race [5, 6, 7].”
The Dataset Nutrition Label: A Framework To Drive Higher Data Quality Standards (2018, background): “Researchers continue to uncover troubling consequences of these systems [2,3].”
Algorithmic Decision Making and the Cost of Fairness (2017, background): “le to determine whether the scores are accurate estimates of p YjX or have been strategically designed to produce 10Some researchers also check whether the AUC of scores is similar across race groups [31]. However, the motivation for examining AUC is not as clear, since the true risk distributions m”
Identifying Significant Predictive Bias in Classifiers (2016, background): “a priori interest, such as race, gender, or income. However, some important subgroups may not be described so simply or be considered a priori. ProPublica’s COMPAS analysis [2] and follow-up analyses [3, 17] focus on predictive bias for subgroups S= fwhite;blackg. In our analysis of COMPAS, we do no”
Ethical Implications and Accountability of Algorithms (2018, unknown): “Critiques of risk assessment or sentencing algorithms have focused on whether the outcome of the algorithm is biased and harms particular groups of individuals (Angwin et al. 2016a; Skeem and Lowenkamp 2015).”
The false promise of risk assessments: epistemic reform and the limits of fairness (2020, background): “Yet this risk assessment would still disproportionately label blacks as “high risk” compared to whites—not because of Human Bias, but because of Population Inequity: due to discrimination and the racialized meaning of “crime” and “risk” [18, 66, 86, 118], African Americans are empirically at higher ”
The Measure and Mismeasure of Fairness: A Critical Review of Fair Machine Learning (2018, background): “When criminologists and practitioners develop or audit a risk assessment tool, they typically check that risk scores are calibrated (Danner et al., 2016; DeMichele et al., 2018; Flores et al., 2016; Skeem and Lowenkamp, 2016).”