Causal Path-specific Analysis
🌸 Coarsening bias from mediator discretization
- Characterize coarsening bias and show that naive discretization induces first-order bias in bin width.
- Propose a bias-reduced functional that evaluates outcome regressions at within-bin conditional means.
🌸 Examining racial disparities in healthcare expenditures
- Introduce novel statistical estimators derived from influence functions.
- Integrate super learning models and apply Two-part model in zero-inflated skewed expenditure data.
- Examine PSEs via each mediator and direct effect.
🌸 Develop R package for flexible causal path-specific analysis
🏷️ - - - - - - - - - - - -
[R package] flexPaths
[papers] Racial Disparities ; Coarsening Bias
[poster] Georgia Statistics Day Conference, 2024
[slides] ACIC 2025; 32nd International Biometric Conference, 2024