Improved mean estimation of sensitive study variable under two phase sampling with subsampling the non-respondents

In this study, we present generalized mean estimators aimed at tackling the challenges associated with non-sampling errors in sensitive survey data. Our approach utilizes both full and optional Randomized Response Technique (RRT) models within a two-phase simple random sampling framework, incorporating information from auxiliary variables. To evaluate the performance of our proposed estimators, we employ a unified measure of efficiency and privacy protection. The validity of these estimators is supported by extensive simulations.