The Social Survey Statistician's Perspective
summary
Surveys used by federal statisticians show concrete limits for DP: randomized response fails at scale because respondents misunderstand and default to denying sensitive items; interviewer and clustering inflate variance and invalidate simple-random-sample formulas; replicate weights and linearization are practical fixes for complex designs; administrative linkage often achieves 80–90% linkage consent in Germany; high-quality face-to-face collection costs are real—face-to-face costs ~$1,000/case—so any privacy noise that shrinks effective sample size forces costly redesigns or larger samples.
AI-written summary · about
chapters (4)
- 0:00Start
- 35:46Effect on Sample Size - Cost (rough estimates)
- 44:59Typical Research Questions
- 53:29Examples of open research questions ....
related talks
55mPANEL: Where Are Statistical Agencies Going?
45mA Practical Method to Reduce Privacy Loss when Disclosing Statistics Based on Small Samples
01h 00mStepping-up: The Census Bureau Sets an Example of How to Be a Good Data Steward in the 21st Century
01h 00mPractical Difficulties in the Application of Differential Privacy for the Release of Large-Scale...
49mStatistical Disclosure Limitation: Where do we go from here?
01h 00mPANEL: Statistical Theory, Privacy and Data Analysis
43mAnalyzing Administrative Data with Privacy Protection in Place
44mAccuracy Disparities and Social Choices in the Design of Privacy Mechanisms