Institution profile

Emory University

Academic institutionnorthamerica · us
Official website
Research library758linked papers
Opportunities0open roles
Selected work

Representative Papers

Difference-in-Differences with a Continuous Treatment

Jul 06, 2021Social Science Research Network

This paper addresses three key challenges in difference-in-differences (DID) estimation under continuous treatment: (i) selection bias due to non-random treatment assignment, (ii) incomparability of treatment effects across varying intensities, and (iii) ambiguous causal interpretation of conventional two-way fixed-effects (TWFE) estimators. We propose a generalized parallel trends assumption and establish the first rigorous identification framework for continuous-treatment DID. We formally prove that TWFE estimators—even in a two-period setting—lack clear causal interpretation under continuous treatment. To overcome this, we develop a bias-corrected, group-weighted estimator grounded in treatment-effect decomposition and augmented with selection-bias sensitivity analysis. Empirically, our method substantially revises policy effect estimates derived from standard TWFE, mitigating systematic misattribution. The proposed approach provides a robust, interpretable tool for causal inference in settings involving graded or intensity-varying interventions.

214 citations14 influentialRead paper

Selection and Parallel Trends

Mar 17, 2022Social Science Research Network

This paper addresses how treatment-group selection threatens the parallel trends assumption in Difference-in-Differences (DiD) estimation—a critical yet under-characterized identification challenge. Method: We formally characterize the empirical content of this threat and derive necessary and sufficient conditions for parallel trends to hold under general selection mechanisms. We propose a “selection-driven bias decomposition framework” that systematically partitions DiD estimation bias into selection effects and time-varying heterogeneity effects, and develop operational benchmarking strategies—both with and without covariates—grounded in causal inference theory, selection modeling, and sensitivity analysis. Contribution/Results: Applied to the National Supported Work (NSW) experiment reanalysis, our approach quantifies and corrects selection bias, substantially improving the credibility of DiD estimates and the robustness of causal conclusions.

27 citations4 influentialRead paper

A Burden Shared is a Burden Halved: A Fairness-Adjusted Approach to Classification

Oct 12, 2021

To address fairness violations in classification arising from imbalanced error rates across protected groups, this paper proposes the Fairness-Adjusted Selection Inference (FASI) framework. FASI introduces False Selection Rate (FSR) control—a concept previously unexplored in fair classification—by provably transforming black-box model outputs into R-values, thereby guaranteeing finite-sample upper bounds on group-wise FSR and achieving statistical parity. The method integrates selection inference, R-value construction, and post-hoc optimization without requiring modifications to the underlying classifier. Experiments on synthetic and real-world datasets demonstrate that FASI substantially reduces inter-group error-rate disparities; FSR control error remains consistently below the prespecified threshold. FASI thus delivers rigorous statistical guarantees, computational efficiency, and strong fairness assurance.

12 citationsRead paper

Jailbreaking and Mitigation of Vulnerabilities in Large Language Models

Oct 20, 2024arXiv.org

Existing research on large language models (LLMs) lacks a unified taxonomy for prompt injection and jailbreaking attacks, and insufficiently evaluates defenses under dynamic, interactive scenarios. Method: We propose the first four-dimensional attack taxonomy—spanning prompt-level, model-level, multimodal, and multilingual pathways—and develop a robust alignment framework tailored to interactive settings, alongside a novel automated jailbreaking detection method. We further conduct systematic defense benchmarking, bias diagnosis of existing evaluation benchmarks, and multi-dimensional security measurement. Contribution/Results: Our analysis reveals critical failure modes of current defenses in dynamic interactions, identifies key research gaps—including ethical implications and data bias—and delivers the first comprehensive technical roadmap for LLM safety alignment.

7 citationsRead paper

Authorship Attribution in the Era of LLMs: Problems, Methodologies, and Challenges

Aug 16, 2024arXiv.org

The rise of large language models (LLMs) has intensified authorship attribution challenges—namely, distinguishing human-authored text from LLM-generated content and resolving ambiguous attribution in human-AI collaborative writing. To address this, we propose the first four-category authorship taxonomy for the LLM era: human-authored, LLM-generated, LLM-attributed, and human-AI collaborative. We systematically survey detection methodologies across four paradigms: statistical features, neural representations, attribution graphs, and prompt engineering—covering models including BERT, RoBERTa, and Llama, as well as watermarking and probability calibration techniques. Further, we introduce a unified evaluation framework balancing cross-domain generalizability and decision interpretability, and establish the field’s first dynamically updated resource repository (llm-authorship.github.io). Our work provides both theoretical foundations and a practical roadmap for enhancing detection accuracy and transparency in LLM-era authorship attribution.

6 citationsRead paper
Recent publications

Latest Papers