My research program investigates how the integration of digital technology and AI into educational systems shapes social stratification. Guided by sociological and educational theories, I use computational methods to examine how institutional policies and practices create or mitigate inequality. My work is organized around two central pillars: 1) uncovering the meso-level mechanisms through which macro-level policies are interpreted and enacted in local practice, and 2) developing robust quantitative tools to ensure inferential making are valid in complex data environments.
I am always excited to connect with other researchers working on these topics. If our interests align and you are interested in potential collaborations, please feel free to reach out via email.
AI, Policy, and Inequality
My current work examines how AI and digital policy diffuse across education systems and how local enactment shapes stratification. Using computational text analysis and network methods, I distinguish between symbolic adoption and substantive implementation of policy. A current project tracks the diffusion of AI education policy in China by integrating multiple data sources, including policy documents and public digital records, to investigate the diffusion and translation dynamics of educational policy from central government to provinces, and to the local school practice.Future studies will explore how teachers’ capacity to engage with AI influences educational outcomes, using a combination of simulation and modeling.
Teachers, Networks, and Culture
My previous research traced how teachers’ online and offline professional networks shape the diffusion of instructional resources, combining digital traces with surveys and administrative records to model networked behavior at scale. This agenda developed into institutional analysis by investigating how institutional structure and teacher agency shape cultural engagement in U.S. classroom content at scale by using NLP methods to score nearly one million lesson plans and assignments on their semantic engagement with racial/ethnic and gender groups.
Methodological Development
This line of work advances statistical and computational tools for valid inference with clustered social data, with an emphasis on diagnosing bias and representing complex dependencies common in education and policy research. Core contributions include a sensitivity analysis for omitted cluster levels in multilevel models, robustness diagnostics for external validity and treatment-effect heterogeneity. I also built a “big‑and‑rich” framework that uses population-representative data to assess selection in digital trace datasets and guide stronger sampling designs.
Publications
A full list of my publications is available in my curriculum vitae.