COMPAS

Learning goals

After completing this assignment, students should be able to:

• Explain how criminal recidivism scores might be racially biased
• Explain why and how a variable like zip code can act as a proxy for race in a predictive algorithm
• Apply, interpret, and compare different measures of algorithmic fairness

Content

• Read:
  • Angwin et al. (2016)
  • Larson et al. (2016) (read at least down to the histograms)
  • ProPublica GitHub repository
• Supplementary reading (optional):
  • Northpointe rebuttal
  • Rudin, Wang, and Coker (2020)
  • Kleinberg, Mullainathan, and Raghavan (2016)
  • O’Neil (2016), Ch. 5 - Civilian Casualties

In-class activity

1. How does Northpointe’s COMPAS algorithm assess risk? What factors does it include? How does race factor into the calculation? Is the algorithm a “black box”?
2. How did ProPublica conduct its investigation? Where did it find the data?
3. Complete worksheet

Next class activity

1. Complete the second worksheet
2. Apply principles of applied ethics to COMPAS

Assignment

Respond to the following prompts on a single piece of paper.

• In one or two paragraphs, briefly weigh the pros and cons of using algorithms like COMPAS. What are the benefits? What are the dangers?

• In one paragraph, reflect on your own feelings about algorithms in the criminal justice space in light of your response to the previous question.

Submission

Hand in before leaving the classroom.

References

Angwin, Julia, Jeff Larson, Surya Mattu, and Lauren Kirchner. 2016. “Machine Bias.” ProPublica; ProPublica. https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing.

Kleinberg, Jon, Sendhil Mullainathan, and Manish Raghavan. 2016. “Inherent Trade-Offs in the Fair Determination of Risk Scores.” arXiv Preprint arXiv:1609.05807. https://arxiv.org/abs/1609.05807/.

Larson, Jeff, Surya Mattu, Lauren Kirchner, and Julia Angwin. 2016. “How We Analyzed the COMPAS Recidivism Algorithm.” ProPublica; ProPublica. https://www.propublica.org/article/how-we-analyzed-the-compas-recidivism-algorithm.

O’Neil, Cathy. 2016. Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy. Crown. https://www.penguinrandomhouse.com/books/241363/weapons-of-math-destruction-by-cathy-oneil/.

Rudin, Cynthia, Caroline Wang, and Beau Coker. 2020. “The Age of Secrecy and Unfairness in Recidivism Prediction.” Harvard Data Science Review 2 (1). https://doi.org/10.1162/99608f92.6ed64b30.