Bayesian semi-supervised inference via a debiased modeling approach
Gözde Sert, Abhishek Chakrabortty, and Anirban Bhattacharya
Econometrics and Statistics, 2025
Winner of ASA Section on Bayesian Statistical Science (SBSS) 2024 Student Paper Award
This work introduces a novel Bayesian semi-supervised inference method for population mean estimation that improves efficiency by modeling specific summary statistics instead of the raw data. The proposed approach utilizes a new Bayesian debiasing technique to produce more robust and efficient estimators, with its effectiveness validated through theoretical results and extensive numerical studies.