Date of Award

Spring 6-4-2023

Document Type

Thesis (Undergraduate)

Department

Quantitative Social Science

First Advisor

Julie Kalish

Second Advisor

Soroush Vosoughi

Abstract

Implicit bias and criminal justice are two concepts that have long been intertwined. There is no justice without neutrality, and yet how do we tell if the actors enforcing the system are actually impartial? In this thesis, I utilize recent advancements in machine learning to attempt to answer this question. Specifically, I use natural language processing to examine the text of opinions written by judges in appellate courts, and I leverage these findings to build quantifiable measures of implicit bias. In particular, I look at the over/under-representation of certain emotions, sentiments and linguistic styles as a proxy for disparate treatment along racial lines. I find that these indicators are not significant predictors for majority opinions, but for dissenting and concurring opinions they are effective proxies for potential judicial cognitive bias.

Download

Share

COinS