My research focuses on designing robust methods for causal inference in complex settings that arise in modern applications, with the same rigor and transparency as in classical designs.
Non-parametric Causal Inference in Dynamic Thresholding Designs
Aditya Ghosh and Stefan Wager
Major revision at the Journal of the Royal Statistical Society Series B: Statistical Methodology
· 2025+
Threshold-based interventions are common in clinical practice, but studying their long-term impact is challenging due to temporal dynamics and carryover effects. We show that a dynamic marginal policy effect at the treatment threshold admits a simple, reduced-form characterization, and develop a local linear regression method for estimation and inference.
Which Covariates to Adjust for? Specification-robust Causal Inference in Observational Studies
Aditya Ghosh and Dominik Rothenhäusler
Major revision at the Annals of Statistics
· 2025+
Covariate adjustment is essential for removing confounding in observational studies, yet researchers often face uncertainty over which covariates to adjust for. Building on debiased machine learning, we establish valid inference for a reweighted target population without committing to a single specification—requiring only that at least one candidate adjustment is valid.
PLRD: Partially Linear Regression Discontinuity Inference
Aditya Ghosh, Guido Imbens, and Stefan Wager
Submitted
· 2025+
Regression discontinuity designs are widely used in applied economics, yet the existing methods often face a trade-off between nominal coverage and interval width. Building on the bias-aware literature, PLRD exploits a partially linear structure and higher-order smoothness to produce intervals that are both valid and short.
Robustness and Efficiency of Rosenbaum's Rank-based Estimator in Randomized Trials: A Design-based Perspective
Aditya Ghosh, Nabarun Deb, Bikram Karmakar, and Bodhisattva Sen
Accepted at Biometrika
· 2026
We study efficiency and robustness of Rosenbaum's rank-based estimator under finite population inference in randomized trials, with and without covariate adjustment.
An Asymptotic Formula for the Chebyshev Theta Function
Aditya Ghosh
Notes on Number Theory and Discrete Mathematics, 25(4), 1–7
· 2019
We derive sharp bounds on the geometric mean of the first n prime numbers.