As a third-year Ph.D. candidate in Statistics at the University of Hong Kong, I research causal inference with a focus on heterogeneous treatment effects (HTE) and integration with resampling frameworks. Under Prof. Stephen M.S. Lee’s mentorship, my work bridges statistical rigor with interdisciplinary applications. My paper in statistics-psychology cross-disciplinary research exemplifies my commitment to expanding causal inference beyond traditional domains. Currently, I am finalizing a novel HTE methodology, presented at EuroCIM 2025 and slated for American Causal Inference SCI2025.

My academic foundation—a first-class BSc in Actuarial Science (HKU) and an MSc in Statistics (University of Michigan, GPA: 3.9/4)—equips me with technical depth and cross-domain adaptability. At Michigan, I represented Prof. Yang Chen and Prof. Shasha Zou’s lab at AGU Fall 2021, applying statistical models to space science challenges.

The Hong Kong Laureate Forum offers a unique platform to amplify the reach of causal inference across humanities, social sciences, and STEM. As a Hong Kong-trained scholar with global exposure, I aim to foster dialogue on statistical innovation’s role in solving real-world heterogeneity. Engaging with laureates and peers will deepen my perspective and inspire collaborative solutions. I am eager to contribute my expertise and learn from this transformative assembly.

The University of Hong Kong

My research focuses on advancing causal inference methodologies, particularly in estimating heterogeneous treatment effects (HTE), by integrating parametric and non-parametric frameworks. At EuroCIM 2025, I introduced a novel "parachuted" estimator for the Conditional Average Treatment Effect (CATE), which achieves double robustness—retaining consistency even if propensity score or outcome models are misspecified. Unlike conventional approaches, our estimator maintains reliability under dual model failures while offering parametric-level efficiency, bridging the gap between robustness and convergence rates.

Methodologically, the estimator synthesizes Augmented Direct Learning (Meng & Qiao, 2022) with kernel-based techniques (Abrevaya et al., 2015), leveraging integration strategies from Lee & Soleymani (2015). We derived its asymptotic distribution, proving alignment with the true CATE and a closed-form variance expression. Additionally, we established bootstrap-based inference validity (Chatterjee & Bose, 2005), enabling reliable uncertainty quantification.

This work extends causal inference’s applicability to complex, real-world heterogeneity in social sciences, space science, and policy evaluation. By unifying resampling frameworks with causal theory, the methodology enhances robustness in scenarios where traditional assumptions falter. At the Hong Kong Laureate Forum, I aim to highlight how such statistical innovations address interdisciplinary challenges, fostering collaborations to refine and deploy these tools across diverse domains.