Statistical Disclosure Control: Past, Present and Future

–
Sponsored by IASS

Instructor

Affiliated with University of Manchester, UK

Prof. Natalie Shlomo

Natalie Shlomo is Professor of Social Statistics in the Social Statistics Department,  School of Social Sciences at the University of Manchester. She is a survey statistician with interests in adaptive survey designs, data linkage and integration, statistical disclosure control and small area estimation. She is an elected member of the International Statistical Institute (ISI), a fellow of the Royal Statistical Society and a fellow of the Academy of Social Sciences.

Course description

The  short course will introduce basic and advanced concepts of statistical disclosure control, privacy and confidentiality. The topics covered include:

  1. introduction to statistical disclosure control, disclosure risk scenarios and types of disclosure risks;
  2. measuring disclosure risk for traditional outputs: microdata and tabular data;
  3. common statistical disclosure control methods and their impact on data quality and utility.

In addition, we introduce a new definition for confidentiality protection called differential privacy which was developed by computer scientists. Differential privacy is a mathematical rigorous definition of a perturbation mechanism that provides formal and quantifiable guarantees of confidentiality. We discuss how differential privacy can be used in the statistical disclosure control tool-kit at statistical agencies as they move towards more advanced, open and flexible modes of data dissemination.

Target audience

For persons  interested in statistical data dissemination, typically based at  statistical agencies and organisations, government agencies, national statistical institutes.

Syllabus

  • Day 1 (3 hours with 10 minute breaks for each hour)

     

    • First Hour: Introduction to statistical disclosure control (SDC), Types of disclosure risks, Disclosure Risk-Data Utility Framework.
    • Second Hour: Disclosure risk assessment for  traditional types of outputs: microdata, tabular data and business data.
    • Third Hour: Some common SDC methods for traditional types of outputs.

Take Home Exercise.

  • Day 2 (3 hours with 10 minute breaks for each hour)

     

    • First Hour: Solution of take home exercise, questions and answers, wrap up from Day 1.
    • Second Hour: Differential Privacy Standard and other  computer science disclosure risk measures.
    • Third Hour: Applications - Flexible Table Builder, New forms of data dissemination.

Required software

None.