Sho Kawano

At JSM 2025 in Nashville

My research mainly focuses on two questions in survey methodology:

  1. How can we produce estimates that are both granular and reliable? For example, estimating the childhood poverty rate in county X from a survey designed for national or state-level estimates.
  2. How can we validate an estimate when there’s no ground truth to check against?

More recently, I’ve found these are the same statistical problem showing up in disaggregated evaluation of AI systems.

Some of my interests include:

  • Small Area Estimation
  • Model Validation
  • AI Evaluation
  • Principled use of Nonprobability Samples
  • Bayesian Modeling

I am co-advised by Zehang Li and Paul Parker.

Publications

  • Sho Kawano, Zehang R. Li, Paul A. Parker. Prediction-Powered Smoothing and Validation for Disaggregated AI Evaluation. Preprint, 2026. arXiv · Code

  • Sho Kawano, Daniel Vendesky, Qianyu Dong, Ethan Pawl, Paul A. Parker, Zehang R. Li, Scott Holan. Nonprobability Samples for Small Area Estimation: A Review and Comparative Simulation Study. Preprint, 2026. arXiv · Code
    • S/O to the big group of UCSC / Mizzou collaborators!
  • Sho Kawano, Paul A. Parker, Zehang R. Li. On Data Thinning for Model Validation in Small Area Estimation. Preprint, 2026. arXiv · Slides · Code
    • This was a really rewarding and fascinating project to work on. Thanks to Ameer Dharamshi for introducing me to data thinning!
  • Sho Kawano, Paul A. Parker, Zehang R. Li. Spatially Selected and Dependent Random Effects for Small Area Estimation with Application to Rent Burden. JRSS-A, 189(3), 2026, pp. 1308–1324. Publication · Slides · Code
    • Winner, Wray Jackson Smith Scholarship (ASA Government and Social Statistics Sections).