Step-by-Step Measurement for an Accurate Study of Society
Egami tests political science claims using statistics, AI scrutiny, and external validity checks to improve civic research confidence.
Key Takeaways
- Naoki Egami studies political methodology, focusing on whether social science findings can be measured accurately and applied beyond their original context.
- He emphasizes external validity, warning that political conclusions drawn from one election, sample, or country may not hold in another setting.
- His work examines AI and machine learning in social science research, especially how to identify and correct new error patterns introduced by AI-generated data.
- Egami argues that rigorous technical methods and substantive political questions must work together to make civic research more reliable and useful.
Naoki Egami Uses Statistics and AI to Test Political Science Findings
Meta description: MIT political scientist Naoki Egami uses statistics, AI scrutiny, and external validity work to help readers judge civic research claims with greater confidence.
Tags: political methodology, AI in research, external validity, MIT
Featured image: abstract friendly illustration of a balance scale, scattered data points, and a ballot-style rectangle, no real people, no logos, no text
Naoki Egami, a political scientist who joined MIT’s Department of Political Science as an associate professor with tenure in 2025, has built a career around political methodology, the field that asks whether studies of voters, campaigns, institutions, and civic life can be measured accurately and applied beyond the setting where they first appeared.
Why external validity matters
Egami studies what social scientists call external validity: whether a finding travels from one sample, election, or country to another. He describes political methodology as a field where researchers ask political questions, then solve them with the tools of applied statistics or computer science. He is also a faculty affiliate of the Statistics and Data Science Center at MIT’s Institute for Data, Systems, and Society.
This matters because civic research can be easy to overread. A study of voters in a lopsided election may not fit a close race. A conclusion about one political system may not fit another. Egami’s work keeps those limits visible.
A path shaped by preparation and opportunity
Egami grew up in Tokyo and attended the University of Tokyo. He liked math and physics, but also political philosophy, and he was unsure how to combine those interests. A workshop about U.S. graduate school changed the direction. He thought it was about MBA programs, but it was about PhD programs and included a political scientist discussing math in the field.
The workshop had 200 people. After it ended, a PhD student panelist found Egami and invited him to a session the next week. He expected another career workshop. Instead, it was an academic seminar with 20 people: 19 professors and Egami, then a first-year undergraduate. There, Kosuke Imai, now at Harvard University, spoke about using statistics in the social sciences. Egami recognized the path he wanted and spoke with Imai.
He later earned his BA after spending a year as an exchange student at the University of Michigan, then applied to graduate schools in the United States and landed at Princeton University. Imai eventually became one of his advisors, along with Rafaela Dancygier, Brandon Stewart, and others. At Princeton, Egami produced papers on external validity and other methodological topics, and audience interest helped him see what the field wanted him to pursue. He earned his PhD from Princeton in 2020, joined Columbia University, and moved to MIT five years later.
Campaign experiments and AI errors
Egami’s research often begins with context. In political science, researchers may gain permission to run field experiments in active campaigns, which can produce strong evidence. But access is uneven. Politicians may let researchers in when they expect to win, while the races that often interest political scientists most are competitive districts, where campaign logic and voter behavior can differ.
He has also scrutinized machine learning and AI tools in social science before the wider AI boom that followed ChatGPT’s rise in late 2022. Some of his work looks for systematic ways to identify errors introduced by AI tools and then build methods that account for those errors.
Traditional social science data is often collected slowly and validated carefully. If AI is used to generate data at scale, new error patterns can appear. Without statistical methods that address those errors, studies may be hard to replicate. Egami argues that researchers need both substantive political questions and rigorous technical methods; otherwise, they may solve real problems only in ad hoc ways.
Recognition and what readers should take from it
Last year, Egami received the Emerging Scholar Award from the Society for Political Methodology. He has also received best paper awards from the American Political Science Association’s sections for political methodology in 2019 and 2025, experimental research in 2024, and political networks in 2022. Honors across three subfields show his range.
He says he tries to start a new research agenda every three to four years, keeping parallel lines of work connected by measurement questions. At MIT, he says he has gained appreciation for an engineering-like habit of working systematically on ongoing problems. He teaches undergraduate and graduate classes and says the people there are genuinely nice.
For readers, the value is simple: a political science finding is not just a number. It is a result produced under particular conditions, using particular tools, and about a particular setting. Egami’s work is useful because it helps readers ask where a claim can safely apply, and where it should be treated with caution.
Comments (0)