A Generalized Negative Binomial Smoothing Model for Sample Disclosure Risk Estimation
We deal with the issue of risk estimation in a sample frequency table to be released by an agency. Risk arises from non-empty sample cells which represent small population cells and from population uniques in particular. Therefore risk estimation requires assessing which of the relevant population cells are indeed small. Various methods have been proposed for this task, and we present a new method in which estimation of a population cell frequency is based on smoothing using a local neighborhood of this cell, that is, cells having similar or close values in all attributes.
The statistical model we use is a generalized Negative Binomial model which subsumes the Poisson and Negative Binomial models. We provide some preliminary results and experiments with this method.
Comparisons of the new approach are made to a method based on Poisson regressionlog-linear hierarchical model, in which inference on a given cell is based on classical models of contingency tables. Such models connect each cell to a ‘neighborhood’ of cells with one or several common attributes, but some other attributes may differ significantly. We also compare to the Argus Negative Binomial method in which inference on a given cell is based only on sampling weights, without learning from any type of ‘neighborhood’ of the given cell and without making use of the structure of the table.
KeywordsRisk Measure Negative Binomial Model Disclosure Risk Work Session Argus Method
Unable to display preview. Download preview PDF.
- 1.Benedetti, R., Franconi, L., Piersimoni, F.: Per-record risk of disclosure in dependent data. In: Proceedings of the Conference on Statistical Data Protection, Lisbon March 1998, European Communities, Luxembourg (1999)Google Scholar
- 2.Bethlehem, J., Keller, W., Pannekoek, J.: Disclosure Control of Microdata. J. Amer. Statist. soc. 85, 38–45 (1990)Google Scholar
- 3.Cameron, A.C., Trivedi, P.K.: Regression analysis of count data, Econometric Society Monographs, vol. 30. Cambridge University Press, Cambridge (1998)Google Scholar
- 4.Di Consiglio, L., Franconi, L., Seri, G.: Assessing individual risk of disclosure: an experiment. In: Proceedings of the Joint ECE/Eurostat Work Session on Statistical Data Confidentiality, Luxemburg, pp. 286–298 (2003)Google Scholar
- 5.Elamir, E., Skinner, C.: Record-level measures of disclosure risk for survey mi- crodata. J. Official Statist. 22 (to appear, 2006)Google Scholar
- 7.Polettini, S., Seri, G.: Guidelines for the protection of social micro-data using individual risk methodology - Application within mu-argus version 3.2. CASC Project Deliverable No. 1.2-D3 (2003), http://neon.vb.cbs.nl/casc/
- 8.Rinott, Y.: On models for statistical disclosure risk estimation. In: Proceedings of the Joint ECE/Eurostat Work Session on Statistical Data Confidentiality, Luxemburg, pp. 275–285 (2003)Google Scholar
- 9.Rinott, Y., Shlomo, N.: A neighborhood regression model for sample disclosure risk estimation. In: Proceedings of the Joint UNECE/Eurostat work session on statis- tical data confidentiality Geneva, Switzerland, pp. 79–87 (2005)Google Scholar
- 10.Rinott, Y., Shlomo, N.: A smoothing model for sample disclosure risk estimation (submitted, 2006)Google Scholar
- 12.Skinner, C., Holmes, D.: Estimating the Re-identification Risk Per Record in Microdata. J. Official Statist. 14, 361–372 (1998)Google Scholar
- 13.Skinner, C., Shlomo, N.: Assessing disclosure risk in microdata using record- level measures. In: Proceedings of the Joint UNECE/Eurostat work session on sta- tistical data confidentiality Geneva, Switzerland, pp. 69–78 (2005)Google Scholar
- 14.Skinner, C., Shlomo, N.: Assessing identification risk in survey microdata using log-linear models (submitted, 2006)Google Scholar