Assistant Professor of Sociology, Toyo University (Tokyo). Quantitative social scientist and methodologist working on causal inference under model uncertainty — which covariates to adjust for, how to report robustness honestly, and what data-driven variable selection does to inference — and on what measurement does to panel-survey data.
- 🌐 sokubo.github.io · ORCID · researchmap · Google Scholar
- 📦
dagmv— causally disciplined multiverse analysis in R (v0.1.3; companion paper arXiv:2609.16618) - 📦
panelcond— refreshment-sample designs for panel conditioning in R - 📄 Preprints: A multiverse of good and bad controls (arXiv:2609.16618) · Optimal covariate adjustment beyond the average treatment effect (arXiv:2609.11222) — replication archives on this account
- 📊 Member of the Japanese Life Course Panel Surveys (JLPS) team, Institute of Social Science, The University of Tokyo
Current projects: candidate-DAG-disciplined multiverse analysis · valid inference after outcome-adaptive selection · estimand-specific optimal adjustment · identification of panel conditioning with refreshment samples · natural experiments in Japanese social policy · dual caregiving across childcare and elder care (JSPS KAKENHI 26K05332).