Noah Dasanaike

Hi, I'm Noah. I'm a fifth-year PhD candidate and the Carl J. Friedrich Fellow in the Department of Government at Harvard University, where I also received my MA (2024) and BA (2022). In the summers of 2025 and 2026, I was a Quantitative Research Intern at Two Sigma, where I continue to work part-time during the academic year.

I am on the 2026–27 academic job market. I expect to complete my PhD in May 2027; my committee is co-chaired by Ryan Enos and Gary King, with Grzegorz Ekiert and Torben Iversen.

My research studies the political economy of cleavages: which social differences become political divisions, why in some places and not others, and why they persist or how they change. I also work on computational methods for large-scale data collection, including the use of language models for historical document digitization, record linkage, and ethnicity inference.

I am the creator of the Small-Area Global Elections (SAGE) Dataset, a database of geocoded, small-area (usually polling-station level) election results for 131 countries. You can read more, download the data, and explore it on the map here.

You can access EnsembleLink, my state-of-the-art method for zero-shot record linkage, here.

At Harvard, I have twice taught the Ph.D.-level "Math Prefresher" and GOV 2020: The Hidden Curriculum (for Gary King). I have also taught STAT 186: Causal Inference (for Susan Murphy) and GOV 97: Political Geography (instructor of record), as well as a GIS workshop for PhD students. Separately, I have also done redistricting consulting.

You can contact me by email at noahdasanaike@g.harvard.edu.

Posts

• How I Digitized 3.4 Million Nazi Party Membership Cards
• What I've Learned From Digitizing 20 Million Historical Documents
• Building SAGE: A Country-by-Country Build Log
• socOCRbench: An OCR Benchmark for Social Science Documents
• namesbench

Research

Job Market Paper

• Why Ethnic Difference Rarely Divides the Vote [abstract]
  Why does politics organize on class in some places and ethnicity in others? Where an ethnic line runs through the local division of labor, a party of class must assemble a coalition across that line, and often cannot. Two configurations make such a coalition difficult. Where one group's counterparties belong to another, an economic complaint is an ethnic one. Where a group's counterparties are its own and serve nobody else, no party can offer it a class appeal at all. Ethnic politics are thus a function of the class party's burden. I assemble more than two billion records of ethnicity, occupation, and residence from twenty-nine countries over two centuries. Of these, only two divide on ethnicity. The theory suggests that class, not ethnicity, is the default state of political divisions, and that ethnic politics are less likely the more ethnic groups there are.

Publications

• Dasanaike, Noah. (2026). The Small-Area Global Elections (SAGE) Dataset. Nature Scientific Data. [article], [dataset & downloads], [code]

Working Papers

Substantive

• Absorbers, Concentrators, and the Urban-Rural Divide revise and resubmit
• Urban-Rural Political Divides Are the Exception, Not the Rule revise and resubmit
• People and Places in the Transmission of Culture: Evidence from German America revise and resubmit
• The Structure of Technological Revolutions: What the Last Structural Transformation Tells Us About the Next One (with Torben Iversen)
• Persistence Without Program: Family Transmission Among Nazi-Émigré Descendants

Methodological

• Probabilistic Race and Ethnicity Prediction Using Group-Specific Name Lists (with Kosuke Imai and Kyla Chasalow)
• Using Embedding Models to Improve Probabilistic Race Prediction (with Kosuke Imai) [working paper]
• Zero-Shot Digitization of Historical Documents with Vision Language Models [working paper]
• Pre-Trained Language Models as Zero-Shot Tools for Social Science Research [working paper] [EnsembleLink] revise and resubmit
• Going Glocal: A Macro-Micro Approach to Political Science (with John Gerring)

Software

• Khanna, Kabir, Brandon Bertelsen, Santiago Olivella, Evan Rosenman, Kosuke Imai, and Noah Dasanaike. "Wru: Who Are You? Bayesian Prediction of Racial Category Using Surname, First Name, Middle Name, and Geolocation." R Package Version 3 (2026).

Additional Work

• Dasanaike, Noah and Grzegorz Ekiert. (2025). Autorytarne Preferencje Wyborców Jako Przyczyna Erozji Demokracji. Almanach 2025/2026, Concilium Civitas. [article pdf]
• Ekiert, Grzegorz and Noah Dasanaike. (2024). The Return of Dictatorship. Journal of Democracy. [article pdf], [publication page], [working paper], [interview]
• Dasanaike, Noah. (2021). Businessperson Deputies and Party Cohesion: Evidence from the Russian State Duma. Party Politics. [article pdf]