Research Areas

Please see Papers for the full list of publications and selected working papers.

1. External validity, Generalization, and Multi-Site Studies

In this research program, I develop methods to improve and assess external validity and generalizability.

I have proposed a formal framework of external validity, methods for site selection in multi-context (multi-site) studies, methods for estimating robustness to external validity bias, and methods to improve external validity of conjoint analysis.

Key words: External validity, Generalization, Cumulative learning

Selected Papers:

• "Designing Multi-Site Studies for External Validity: Site Selection via Synthetic Purposive Sampling."
  Naoki Egami and Diana Da In Lee
  [R package]

• "Quantifying Robustness to External Validity Bias."
  Martin Devaux and Naoki Egami
  

• "Elements of External Validity: Framework, Design, and Analysis."
  Naoki Egami and Erin Hartman
  American Political Science Review (2023)
  

• "Improving the External Validity of Conjoint Analysis: The Essential Role of Profile Distribution."
  Brandon de la Cuesta, Naoki Egami, and Kosuke Imai
  Political Analysis (2022)
  

• "Covariate Selection for Generalizing Experimental Results: Application to A Large-Scale Development Program in Uganda."
  Naoki Egami and Erin Hartman
  Journal of the Royal Statistical Society: Series A (Statistics in Society) (2021)
  

2. Causal Inference with Network and Spatial Data

In this research program, I develop methods to identify and estimate causal effects when units interact with each other on networks and across space.

I have worked on identification of causal network effects from observational data in the presence of unmeasured confounding as well as from randomized experiments with interference.

Key words: Interference, Peer effects, Diffusion effects, Homophily

Selected Papers:

• "Identification and Estimation of Causal Peer Effects Using Double Negative Controls for Unmeasured Network Confounding."
  Naoki Egami and Eric J. Tchetgen Tchetgen
  Journal of the Royal Statistical Society: Series B (Statistical Methodology) (Forthcoming)
  - Winner of the Best Conference Paper Award in Political Networks.
  

• "Identification of Causal Diffusion Effects Using Placebo Outcomes Under Structural Stationarity."
  Naoki Egami
  Journal of the Royal Statistical Society: Series A (Statistics in Society) (Forthcoming)
  - Winner of the Gosnell Prize for Excellence in Political Methodology.
  

• "Spillover Effects in the Presence of Unobserved Networks."
  Naoki Egami
  Political Analysis (2021)
  - Winner of the John T. Williams Dissertation Prize.
  

3. Machine Learning for the Social Sciences and Causal Inference

In this research program, I extend machine learning methods, which are originally developed for prediction, to answer causal questions in the social sciences.

I have worked on how to make causal inference with text data, and extended regularized regression to estimate causal interaction in factorial experiments.

Key words: Machine learning, Text analysis, Regularization

Selected Papers:

• "Using Imperfect Surrogates for Downstream Inference: Design-based Supervised Learning for Social Science Applications of Large Language Models."
  Naoki Egami, Musashi Hinck, Brandon M. Stewart, and Hanying Wei
  Advances in Neural Information Processing Systems (NeurIPS) (2023)
  

• "How to Make Causal Inferences Using Texts."
  Naoki Egami, Christian J. Fong, Justin Grimmer, Margaret E. Roberts, and Brandon M. Stewart
  Science Advances (2022)
  

• "Causal Interaction in Factorial Experiments: Application to Conjoint Analysis."
  Naoki Egami and Kosuke Imai
  Journal of the American Statistical Association (2019)