The Microdata Analysis System at the U.S. Census Bureau

• Activity Number: 338
• Type: Topic Contributed
• Date/Time: Tuesday, August 2, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Michael Freiman*+ and Michael Freiman and Laura Zayatz and Lisa Singh and Jiashen You and Michael DePersio
• Companies: U.S. Census Bureau and U.S. Census Bureau and U.S. Census Bureau and U.S. Census Bureau and University of California at Los Angeles and U.S. Census Bureau
• Keywords: Data Confidentiality ; Remote Access Servers ; Universe Subsampling ; Synthetic Data ; Regression ; Disclosure

Abstract

The U.S. Census Bureau has the responsibility to release high quality data products while maintaining the confidentiality promised to all respondents under Title 13 of the U.S. Code. This paper describes a Microdata Analysis System (MAS) that is currently under development, which will allow users to receive certain statistical analyses of Census Bureau data-such as cross-tabulations, regressions (with diagnostic plots), histograms and scatterplots-without ever having access to the data themselves. Such analyses must satisfy several statistical confidentiality constraints; those that fail these constraints will not be output to the user. In addition, all analyses are performed after application of the Drop q Rule, a data subsampling routine, and regressions involving categorical predictors sometimes require modification. We describe the system's capabilities, as well as the confidentiality protections and the major types of attacks they prevent, then conclude with a description of other approaches to creating a system of this sort, and some directions for future research.