A two-stage randomized response technique for simultaneous estimation of sensitivity and truthfulness

Area of Science:

• Statistics
• Survey Methodology
• Data Privacy

Background:

• Conventional randomized response (RR) models offer insufficient privacy for sensitive survey questions.
• Estimating key parameters like truthfulness (T) and success probability (π) is challenging with existing RR models.
• Existing RR methods often compromise respondent secrecy and data accuracy.

Purpose of the Study:

• To propose an improved RR technique offering enhanced privacy protection.
• To enable simultaneous estimation of truthfulness (T) and success probability (π).
• To adapt the technique for both simple and stratified random sampling designs.

Main Methods:

• A two-stage randomization process is employed, building on prior RR research.
• The model is analytically derived and evaluated for computational efficiency and algebraic properties.
• Stratified random sampling is incorporated for complex survey designs.

Main Results:

• The proposed model demonstrates superior privacy protection and parameter estimation accuracy compared to traditional methods.
• Simulations show reduced variance and enhanced privacy under both simple and stratified sampling.
• The method improves respondent participation and data quality without sacrificing statistical accuracy.

Conclusions:

• The enhanced RR technique provides a statistically sound and effective solution for surveys on sensitive topics.
• It guarantees improved data quality and robust respondent protection.
• The model is applicable to diverse survey designs, enhancing reliability and participation.