Better Bargaining: Using Machine Learning to Reduce Sentencing Disparities

poster of research in article

Students

2020
Wharton

Faculty

Professor of Law, Business Economics, and Public Policy

Project Summary

This summer, I worked with Professor David Abrams and his research team. Our goal was to evaluate whether more information given to public defenders reduces sentence disparities and if this greater information provision leads to better outcomes. More specifically, this additional information comes in the form of predictions on the rates of incarceration and sentence lengths of felonies. Using a machine learning algorithm, we develop these predictions based on similar past cases. These predictions are defendant-specific depending on his or her offense, criminal history, age, and judge.

In the Las Vegas Public Defender Office, we are running a year-long pilot in which we provide these predictions to randomly assigned attorneys and cases. The tool is accessed via a password-protected website that public defenders can access. We developed the tool to specifically aid these attorneys during the plea bargaining process. If successful, this could substantially decrease inequalities in the criminal justice system across the country.

Although I had no prior experience in the law field, this research opportunity allowed me to learn a great deal about the American justice system. I would not have had the fortune to gain this knowledge if it were not for this project. Furthermore, I learned about the challenging and arduous research process. With this first-hand experience, I have a better understanding of what exactly research entails and the process by which it develops. Now, if I wish to pursue a long-term academic career in the future, I will already have a good understanding of where to begin.