Research

My research uses data science methods to explain and prescribe social and political problems. For twenty-five years that has meant electoral behavior and democratization in East Asia, measured with whatever instrument political conditions allow — probability samples where they are possible, and generative methods where they are not.

Synthetic data and generative methods

My most active line. It brings together silicon sampling with large language models to generate synthetic survey respondents; Bayesian networks and directed acyclic graphs with do-calculus to identify media effects where experiments are impossible; Wasserstein GANs for modeling political support dynamics; and ROC-based out-of-sample validation as the discipline on all of it.

The problem I want to solve is validation. The field is producing synthetic respondents faster than it is agreeing how to distinguish a good one from a bad one. I am developing a four-part protocol — out-of-sample ROC, distributional distance on the joint rather than the marginals, subgroup recovery, and known-answer backtesting against historical elections — with an accompanying open R package.

Measurement when political space closes

As Co-Principal Investigator of the Hong Kong Election Study I helped run an annual probability survey from 2015 to 2023, spanning a period in which the territory moved from contested electoral politics to a substantially closed political system. The methodological claim running through this work is that standard attitudinal instruments — internal political efficacy above all — do not carry fixed meaning across regime types.

Media effects and generational cleavage

Taiwan is an unusually clean setting for studying partisan media effects, because channel partisanship is explicit and recognized by viewers. My work incorporates television viewership directly as a predictor of party support across the KMT, DPP and TPP, and examines generational structure in those effects.

AI readiness in governance

Applied work on AI readiness in public governance and digital democracy in Taiwan, and on generative methods for integrating public preferences into carbon management and energy policy.

Research projects

• Co-Principal Investigator, Hong Kong Election Study — annual survey, 2015–2023
• Advisor, COMETrends — annual national survey, 2020–
• Member, North Texas Quality of Life project
• Co-Investigator, Higher Education Research Survey

Conference papers and invited lectures are listed on the talks page.

Peer review

American Political Science Review · American Journal of Political Science · British Journal of Political Science · Comparative Political Studies · Political Analysis · Political Behavior · Public Opinion Quarterly · International Studies Quarterly · Social Science Quarterly · National Science Foundation