Noise infusion as a confidentiality protection measure for graph-based statistics

Article type: Research Article

Authors: Abowd, John M.^{a; *} | McKinney, Kevin L.^b

Affiliations: [a] Cornell University, Ithaca, NY, USA | [b] U.S. Census Bureau, USA

Correspondence: [*] Corresponding author: John M. Abowd, 261 Ives Hall, Cornell University, Ithaca, NY 14853, USA. Tel.: +1 607 255 8024; E-mail:john.abowd@cornell.edu

Abstract: We use the bipartite graph representation of longitudinally linked employer-employee data, and the associated projections onto the employer and employee nodes, respectively, to characterize the set of potential statistical summaries that the trusted custodian might produce. We consider noise infusion as the primary confidentiality protection method. We show that a relatively straightforward extension of the dynamic noise-infusion method used in the U.S. Census Bureau's Quarterly Workforce Indicators can be adapted to provide the same confidentiality guarantees for the graph-based statistics: all inputs have been modified by a minimum percentage deviation (i.e., no actual respondent data are used) and, as the number of entities contributing to a particular statistic increases, the accuracy of that statistic approaches the unprotected value. Our method also ensures that the protected statistics will be identical in all releases based on the same inputs.

Keywords: Confidential disclosure, noise-infusion, graph theory

DOI: 10.3233/SJI-160958

Journal: Statistical Journal of the IAOS, vol. 32, no. 1, pp. 127-135, 2016