Compas

COMPAS (Correctional Offender Management Profiling for Alternative Sanctions) is a risk assessment algorithm used in the U.S. criminal justice system. It is known for its evaluation of defendants’ likelihood of reoffending. It was developed by Northpointe, now Equivant, a private company based in Ohio, in the 1990s. COMPAS includes a questionnaire covering 137 factors, including criminal history, personal behavior, and social environment. Algorithms then process these factors to assign a risk score ranging from one to ten. COMPAS is instrumental in determining the severity of prison sentences, the likelihood of an offender reoffending, the potential for a criminal to commit crimes during parole or release, and the probability of the defendant committing violent crimes. The tool is in use in New York, Wisconsin, California, and parts of Florida, where it aids judges in making crucial decisions about parole or the severity of punishment.

According to a study by ProPublica in 2016, COMPAS’ risk assessments were nearly twice as likely as white defendants to be incorrectly labeled as high risk for reoffending, while white defendants were more often mislabeled as low risk. Northpoint argued that its assessments were unbiased and fair. The study led to a discussion about the effectiveness of tools like COMPAS. Opponents argued that COMPAS is just continuing systemic racism.

Background

The United States has the highest amount of prisoners in the world.

According to data from the 2022 Census and the Bureau of Justice, Black individuals comprised 13.7% of the general population but 32% of the prison population. Similarly, Hispanic individuals made up 19.5% of the general population but represented 23% of those incarcerated. White individuals, who account for 58.4% of the general population, are just 31% of the prison population.

Questions.

1. Is Compas an effective tool for risk assessment? Does the tool have any biases?
2. Should Compas’s algorithm and its inner workings be disclosed to ensure transparency?
3. Suggest (if you think Compas is ineffective) alternative methods to assist judges in making decisions. Use the following database to find alternative solutions: https://criminaljustice.tooltrack.org/
4. How can algorithms get rid of racial biases?

Resources:

Angwin, J., Larson, J., Mattu, S., & Kirchner, L. (2016, May 23). How we analyzed the COMPAS recidivism algorithm. ProPublica. https://www.propublica.org/article/how-we-analyzed-the-compas-recidivism-algorithm

Grgic-Hlaca, N., & Redmiles, E. M. (2024). Code is law: How COMPAS affects the way the judiciary handles the risk of recidivism. Artificial Intelligence and Law, 32(1), 1-23. https://link.springer.com/article/10.1007/s10506-024-09389-8

Thomas, S. (2023). The fairness fallacy: Northpointe and the COMPAS recidivism prediction algorithm (Bachelor’s thesis). Columbia University. https://academiccommons.columbia.edu/doi/10.7916/ab13-jf83

Veale, M., & Brass, I. (2020). Uncertainty, risk, and the use of algorithms in policy decisions: A case study on criminal justice in the USA. Policy Sciences, 53(4), 635-658. https://link.springer.com/article/10.1007/s11077-020-09414-y