Digital Government:
Project Update - October, 2001
1. ROS SPECIAL ISSUE. One final reminder! The upcoming special issue of Research in Official Statistics on Bayesian methods in official statistics will contain three papers arising from the project:
- Fienberg, S. E., and Makov, U. E.: Uniqueness and Disclosure Risk:
Urn Models and Simulation
- Lee, J., Holloman, C., Karr, A. F., and Sanil, A. P.: Analysis of Aggregated
Data in Survey Sampling with Application to Fertilizer/Pesticide Usage
Surveys
- Trottini, M.: A Decision-Theoretic Approach to Data Disclosure Problems
All are available at www.niss.org/dg/technicalreports.html
2. NEW THREADS OF RESEARCH.
Three new threads of research are being initiated, each led by one of the three NISS postdocs assigned to the project:
- OPTIMAL TABULAR RELEASES. Complementing work on (dynamic) table servers, methodology and algorithms are being developed for producing optimally informative fixed sets of releases (of marginal subtables of a large table), subject to disclosure risk constraints. Adrian Dobra is playing the lead role.
-
DATA SWAPPING. Studies of data swapping in the context of NCES data are meant to elucidate the sensitivity of risk measures to parameters of swapping algorithms (such as swapping rates) as well as algorithms themselves. The approach will be heavily computational and Bayesian. Shanti Gomatam is leading this effort.
-
AGGREGATION. Using BLS data, further studies of aggregation will address aggregation in multiple variables simultaneously, complex disclosure criteria and statistical consequences of aggregation. Ashish Sanil leads this component of the project.
3. NEW PUBLICATIONS.
-
Fienberg, S. E., Makov, U. E., Meyer, M. M., and Steele, R. J. (2001). "Computing the exact distribution for a multi-way contingency table conditional on its marginal totals," In A. K. M. E. Saleh, ed., _Data Analysis from Statistical Foundations: Papers in Honor of the 75th Birthday of D. A. S. Fraser_, Nova Science Publishing, 145-165.
4. PRESENTATIONS
At Carnegie Mellon University
-
George Duncan: Disclosure Risk and Data Utility: The R-U Confidentiality Map (October 1, 2001)
-
** George Duncan: How is IT Transforming Privacy? (October 16, 2001)
At the Department of Statistics, University of Chicago
-
Stephen E. Fienberg: Conditional Distributions, Log-linear Models, and Disclosure Limitation" Methods (October 22, 2001)
Prepared but not given
-
** Sallie Keller-McNulty (with George Duncan and Lynne Stokes): Disclosure Risk vs. Data Utility: The R-U Confidentiality Map
Copies of those marked with ** are available at http://www.niss.org/dg/presentations.html
5. CONGRATULATIONS! Adrian Dobra successfully defended his dissertation "Statistical Tools for Disclosure Limitation in Multi-way Contingency Tables" on October 26. Adrian has now completed all requirements for his Ph.D. in statistics at Carnegie Mellon University, under the direction of Stephen Fienberg.
