Through the Master of Science in Political Analytics program’s capstone course, students tackle complex, real-world political questions while acting as data-driven consultants. These projects—students’ culminating output of the program—translate advanced analytics into practical political insight across campaigns, public policy, civic engagement, and democratic participation.
The fall 2025 capstone presentations were no exception. The cohort’s final projects explored representation in political surveys, experimental public opinion research, structural barriers to voter turnout, the civic and electoral impact of Minority Serving Institutions, and the use of automated tools to guide campaign messaging strategy.
Spanning topics from local policy debates to national electoral dynamics, the presentations demonstrate how data science can inform both democratic participation and political decision making.
See their work in action:
Political Analytics student Kayla Willis
Who’s Missing from Political Surveys? Revealing Age Gaps in Election Cycles Between 2014 and 2020
Aminah Al-Jaber and Marina Ceserano examined whether young adults are systematically underrepresented in U.S. political surveys and whether that bias changes between midterm and presidential elections. Using Cooperative Congressional Election Study (CCES) data benchmarked against American Community Survey (ACS) population data, they calculated representation bias across five age groups and four election years and found consistent underrepresentation of 18–24-year-olds, with gaps largest during the midterms, and narrowing, but not disappearing, in presidential years. To prevent distorted perceptions of the electorate and policy preferences, their study recommends methodological improvements such as multi-mode outreach, targeted oversampling, and enhanced weighting to ensure surveys more accurately reflect the full democratic population.
Watch their presentation here.
Yea or Neigh? Public Opinion on NYC’s Proposed Horse-Drawn Carriage Ban
Kelly Sexton’s capstone project examined how different information frames influence support for banning horse-drawn carriages. Through a randomized survey experiment of 2,000 New York City residents, she found that while baseline support for the ban stood at 55%, 80% of respondents prefer compromise solutions that protect both workers and animals, only 10% support the ban as written, and 76% would consider an independent veterinary oversight panel as an alternative. Sexton tested five conditions: control, immigrant worker impact, real estate interests, policy alternative (independent veterinary oversight), and a combined frame, using a pre/post design. Her findings revealed that public opinion is highly responsive to information, that voters reject an “animals vs. workers” narrative, and that a policy path centered on oversight and worker protections has substantially broader support than an outright ban.
Watch the full presentation here.
The Role of Food Insecurity in Predicting Political Participation in NYC
Matthew Lee and Felicitas Hadad investigated whether incorporating a tract-level Food Insecurity Index improves the accuracy and explanatory power of voter turnout predictions in New York City. Using USDA Food Access Research Atlas data, ACS population data, and 2024 New York State voter file data, Lee and Hadad constructed a standardized index based on poverty rates, SNAP reliance, and distance from supermarkets, then integrated it into a logistic regression model alongside demographic, socioeconomic, and party controls. What they found was a strong negative relationship between food insecurity and turnout. Their study also concluded that food insecurity is not only a public health and economic concern but also a meaningful structural barrier to democratic participation.
See the full presentation here.
Political Analytics student Marie Pesch
The Role of Minority Serving Institutions in Democratic Vote Share and Public Service
Taylor Cook and Kayla Willis examined how Minority Serving Institutions (MSIs) influence Democratic presidential vote share and civic leadership across the United States. Using NASA’s MUREP dataset of more than 800 MSIs, county-level election data from 2020 and 2024, and regression and spatial analysis, their study tested whether graduation rates and minority student representation predict Democratic vote share. The findings revealed that minority representation within MSIs is the strongest and most statistically significant predictor, while graduation rates have only a weak effect. They concluded that MSIs function as political and civic anchors, particularly in medium to high cluster states, and recommended increased strategic investment and partnership building to strengthen democratic engagement and public service pipelines.
Watch their presentation here.
Testing A New Kind of DNA
Bryon Vaughn, Chisom Adubasim, and Marie Pesch’s capstone project, introduced an automated tool designed to help political campaigns determine when to pivot messaging in response to shifts in news coverage during the 2025 New York City mayoral election. DNA—Dynamic Narrative Assistant—ingests daily articles from New York-based outlets, assigns topics, evaluates sentiment and framing using language models, and weighs coverage volume and stakeholder impact to issue a daily pivot or hold decision along with a rationale and draft messaging. The system was tested on simulated campaign environments for Andrew Cuomo and Zohran Mamdani and evaluated against human raters using a structured rubric. Results showed stronger alignment for Cuomo, with higher agreement and performance metrics, while performance for Mamdani was weaker under default settings. After fine-tuning campaign-specific parameters, accuracy improved substantially for both candidates. The project highlighted the promise of collaborative intelligence between automated systems and human strategists and suggests broader applications beyond electoral politics.
See their presentation here.
About the Program
The Columbia University M.S. in Political Analytics program provides students quantitative skills in an explicitly political context, facilitating crosswalk with nontechnical professionals and decision-makers—and empowers students to become decision-makers themselves.
The 36-credit program is available part-time and full-time, on-campus. Learn more about the program here.
For general information and admissions questions, please call 212-854-9666 or email politicalanalytics [[at]] sps [[dot]] columbia [[dot]] edu.
