Sharing Precise Geographies in Public Use Data

• Activity Number: 304
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Government Statistics
• Author(s): Hao Wang* and Jerome P. Reiter+
• Companies: Duke University and Duke University
• Keywords: Confidentiality ; Disclosure ; Geography ; Multiple Imputation ; Spatial Analysis

Abstract

When statistical agencies release data to the public, they are obligated by law to protect the confidentiality of survey respondents' identities. Geographical variables are among the most challenging data to share when making a primary data source available to public. In this paper, we develop a multiple imputation approach for protecting the confidentiality of data with geographic identifiers. We propose the use of a sequence of classification and regression trees to generate synthetic geography that can preserve spatial associations. We then describe a novel disclosure risk measure in the multiply-imputed data sets, accounting for released information from the multiple datasets. Finally, we illustrate the computation using the Durham city mortality data set.