Study to Alter the Nuisance Effect of Non-Response Using Scrambled Mechanism

Area of Science:

• Statistics
• Survey Methodology
• Biometrics

Background:

• Biometric sample surveys require methods to collect data on sensitive issues for policy planning.
• Randomized response/scrambled response techniques are employed for sensitive data collection.
• There is a need for scrambled response techniques applicable to successive occasions for time-varying sensitive issues.

Purpose of the Study:

• To propose new effective scrambled response models for estimating population means of quantitative sensitive characteristics.
• To evaluate the efficacy of proposed models using degree of privacy protection and unified measure approaches.
• To explore the utility of proposed models in successive sampling scenarios.

Main Methods:

• Developed new additive and multiplicative scrambled response models.
• Assessed model efficacy using privacy protection and unified measure metrics.
• Utilized MATLAB for efficiency checks, empirical, and simulation studies.
• Explored model utility in successive sampling using exponential-type estimators.

Main Results:

• Proposed models demonstrated percent relative efficiencies greater than 100% compared to the Bar-Lev et al model.
• Privacy protection values for proposed models were generally above 0.5 and closer to 1.
• Proposed models showed smaller values in the unified measure approach compared to the Bar-Lev et al model.
• Successive sampling application resulted in percent relative efficiencies consistently over 100% for proposed models.

Conclusions:

• The proposed scrambled response models outperform existing models in estimating sensitive characteristics.
• These models are suitable for human surveys addressing highly sensitive issues.
• The models are effective in successive sampling, offering a robust alternative for time-varying sensitive data.