A New Data Collection Technique for Preserving Privacy.

Area of Science:

• Computer Science
• Statistics
• Social Science

Background:

• Data collection for research is hindered by privacy concerns.
• Traditional methods often compromise data reliability or participant privacy.
• Web-based and mobile data collection are increasingly common.

Purpose of the Study:

• To introduce a novel data collection method, triple matrix-masking (TM2), for enhanced privacy protection.
• To ensure that masked data remains analyzable using standard statistical methods.
• To improve upon existing privacy-preserving techniques like Warner's randomized response.

Main Methods:

• The TM2 method employs sequential matrix transformations for data masking.
• Separate key management prevents any single entity from accessing raw data.
• A masking service provider and data collectors perform distinct masking steps.
• Includes an error-checking mechanism and partial masking for selective data release.

Main Results:

• TM2 offers strong privacy protection, preventing researchers from viewing original data.
• Statistical inference on binary and normal data yields identical results to using original data.
• No efficiency loss in statistical inference is observed compared to unmasked data.

Conclusions:

• The TM2 method effectively balances data privacy and analytical utility.
• It provides a robust solution for collecting sensitive information in medical and social research.
• Enhanced features like error checking and partial masking increase the method's applicability and trustworthiness.