About
Professional Summary
I am a fourth-year PhD student in Biostatistics at the Yale School of Public Health. My research lies at the intersection of causal inference, econometrics, and machine learning, with a focus on developing rigorous and practical methods for data-driven decision-making. My current interests include treatment effect heterogeneity, policy learning, causal mediation analysis, principal stratification, and instrumental variable methods. I am particularly motivated by applications in precision medicine, clinical trials, observational studies, and social experiments.
Latest
News
New preprint
Causal mediation and treatment effect heterogeneity
Nonparametric heterogeneous causal mediation with orthogonal machine learning
In recent work, we develop a class of weighted orthogonal causal machine learners for estimating how causal mediation effects vary across individual characteristics, together with flexible estimation procedures and pointwise and uniform confidence bands. To improve finite-sample performance, we further combine orthogonal learning with targeted learning. Applications in cardiometabolic health, psychology, and education illustrate heterogeneity in mediated effects across individual profiles.
New preprint
Policy learning
Policy learning and individualized treatment rules
In recent work, we introduce orthogonal double residual learning (ODRL), a flexible
causal machine learning framework that directly learns optimal individualized
treatment rules through cost-sensitive classification based on the product of
treatment and outcome residuals. The proposed method is robust to nuisance
estimation error and poor overlap. The accompanying odrlITR R package
implements the method.
Focus
Research Areas
Causal Inference
Rigorous methods for identification and estimation of causal effects.
Econometrics
Econometric tools for identification, estimation, and policy evaluation.
Treatment Effect Heterogeneity and Policy Learning
Methods for estimating heterogeneous effects and learning data-driven treatment policies.
Principal Stratification and Mediation
Frameworks for post-treatment events, mediation, and interpretable causal pathways.
Training
Education and Training
PhD Student in Biostatistics
Current
Yale School of Public Health
Research Mentorship
Advisor: Prof. Fan Li
Department of Biostatistics, Yale
Master's Research
Clinical trial design
Sample size estimation
Writing
Selected Publications
In press
Principal stratification
Semiparametric Principal Stratification Analysis Beyond Monotonicity
Tong J, Kahan B, Harhay MO, Li F. Statistica Sinica, in press.
2026
2025
2025
2024
2023
2023
Current Work
Preprints
2026
2026
Individualized treatment rules
Orthogonal double residual learning for optimal individualized treatment rules
Tong J, Li F. arXiv preprint arXiv:2608.24085.
2026
Treatment heterogeneity
Doubly cross-fit debiased machine learning of heterogeneous treatment effects under principal stratification
Tong J, Li F. arXiv preprint arXiv:2606.29076.
2026
Causal mediation
Causal Mediation in Cluster-Randomized Trials With Multiple Mediators: Spillover-Aware Decomposition, Identification, and Semiparametric Efficient Inference
Tong J, Cheng C, Li F. arXiv preprint arXiv:2604.10710.
2025
Causal estimands
On the permutation equivariance principle for causal estimands
Tong J, Li F. Major revision at Statistica Sinica. arXiv:2510.11863.
Open Source
Software & R Packages
R package
odrlITR
Orthogonal Double Residual Learning for Optimal Individualized Treatment Rules
Learns individualized treatment rules using cross-fitted nuisance estimation and the double residual score, with tree, linear, SVM, and ReLU policy learners.
pak::pak("deckardt98/odrlITR")
R package
PSor
Semiparametric Principal Stratification Analysis Beyond Monotonicity
Estimates principal causal effects under sensitivity models that relax monotonicity, with doubly robust and debiased machine-learning inference.
install.packages("PSor")
R package
MRStdCRT
Model-Robust Standardization in Cluster-Randomized Trials
Estimates marginal treatment effects using model-robust standardization for cluster-randomized trials.
install.packages("MRStdCRT")
Connect