Kate Daugherty utilizes machine learning to analyze complex issues in social science

April 10, 2023

Kate Daugherty, age 24, Class of 2023


Daugherty is majoring in sociology, while pursuing the undergraduate certificate from the Center for Statistics and Machine Learning (CSML).

Undergraduate Work:

Daugherty is interested in utilizing machine learning tools to uncover hidden trends in social science, a track that she had focused on for her CSML independent project, which was done in her junior year. Daugherty looked at how bail is set in Seattle Municipal Court and used quantitative analysis to see if there were any biases impacting bail decisions, particularly regarding prosecutors’ influence.

“Although sociological research has found racial bias in bail assignments, prosecutorial influence on a judge's bail decision remains largely unexplored,” said Daugherty.

For her project, Daugherty took observational data from court watchers who sat on arraignment hearings, which occur 24 to 48 hours after a defendant’s arrest. Judges use the meeting to read the police report and set conditions of release, such as bail amount. Daugherty participated as one of the court watchers as part of Court Watch WA, a nonprofit group that sends volunteers to observe courts in Seattle.

Daugherty then developed three multivariate linear regressions to explore any factors that may influence judges’ bail decisions. Variables included in her study were the bail recommendation from prosecutors and various characteristics of the defendant that may provoke bias, such as race or ethnicity.

Previous studies have shown that courts are rife with racial bias, with Black defendants more likely to get assigned bail and at higher amounts than white defendants, Daugherty said.

“These studies analyzed court records, which hold information on defendant characteristics and judges’ bail decisions but crucially do not preserve the prosecutors’ bail request,” said Daugherty, who called her project the first analysis of prosecutorial influence on bail setting.

Her analysis found that a judge's bail decision is strongly associated with the prosecutor's recommendation, even if two defendants have a similar background and prior arrest records. For example, two defendants with similar profiles go in front of a judge, and one prosecutor calls for $5,000 bail for one defendant, and another prosecutor calls for $15,000 for the second defendant. The judge will follow the prosecutor’s lead and allow for higher bail. 

“At a 1% significance level, my analysis shows that, when controlling for other factors, judges assign an additional $.80 in bail for every $1 requested by a prosecutor,” she said. “I find that heterogeneity in judicial bail assignments can be attributed to prosecutor recommendations at a statistically significant level. These findings are striking, as they suggest that the prosecutor's request influences the judge's decision regardless of the facts of the case.”

Daugherty continued, “And this suggests that if prosecutors are the ones with bias, we could be unfairly blaming judges, and we might be misallocating resources for anti-bias training or other interventions.”

For her senior thesis, which also has a machine learning component, Daugherty is looking at a national survey on guns in which people are asked about their experience with gun violence, approval of gun control laws, and other questions. The survey is from the University of Chicago’s Crime Lab, where she has served as a research assistant. She built a random forest for each of the policy questions that is asked to figure out any correlations between policy answers and a responder’s background. Daugherty is complementing that study by interviewing people at gun ranges in New Jersey.

After her graduation, Daugherty will continue working as a data analyst at the Crime Lab because she admires their work on “high level, quantitative, empirical research on violence.”

“I'm really excited to be doing something that feels meaningful, that I'm excited for, and that Princeton has specifically trained me to do,” said Daugherty, crediting the CSML certificate for preparing her for this work.

Extracurricular activities:

Daugherty interned at Princeton’s Gender + Sexuality Resource Center and was a volunteer “peer” at the Sexual Harassment/Assault Advising, Resources and Education (SHARE) office on campus. As a peer, she planned programming on campus and served as a first responder to students experiencing assault or interpersonal harm.

For Fun:

Daugherty enjoys playing board games and puzzles. She also enjoys hiking in the Pacific Northwest.