Calibrated-Two Optional Randomized Response Techniques (C-TORRT) for the estimation of quantitative sensitive...

Area of Science:

• Survey Methodology
• Statistical Inference
• Data Privacy

Background:

• Accurate estimation of sensitive quantitative variables is difficult due to respondent reluctance.
• Existing randomized response techniques (RRT) often lack efficiency and robustness.
• There is a need for improved methods balancing privacy and data accuracy.

Purpose of the Study:

• To propose novel Calibrated-Two Optional Randomized Response Techniques (C-TORRT).
• To enhance estimation accuracy and respondent privacy using auxiliary information.
• To evaluate the theoretical and empirical performance of C-TORRT models.

Main Methods:

• Development of new C-TORRT models incorporating calibration methods.
• Theoretical analysis demonstrating unbiasedness, reduced variance, and enhanced privacy.
• Empirical validation using simulated and real-life data (academic records).

Main Results:

• C-TORRT models showed significantly lower variance and higher percentage relative efficiency (PRE) compared to existing RRT models.
• Empirical studies demonstrated superior performance in both simulated populations and real-life data applications.
• Proposed models achieved a better balance between efficiency and privacy protection.

Conclusions:

• The proposed C-TORRT models offer superior efficiency, precision, and privacy protection.
• C-TORRT provides a robust alternative for collecting sensitive quantitative data.
• These findings advance survey methodology for sensitive attribute estimation.