Laura Zayatz

Essentials of the disclosure review process: a federal perspective.

Many researchers need to make arrangements to share de-identified electronic data files. However, the ways in which respondent identity may be protected are not well understood or are assumed to be the special province of large statistical agencies or specialized statisticians. Approaches to data sharing and protecting respondent identity have been pioneered by federal agencies which gather data vital to political and economic decision making. These agencies are required by statutory law both to assure confidentiality and to share data in usable form with other governmental agencies and with scholars who perform needed analyses of those data. The basic principles of disclosure limitation developed by the Census Bureau, the National Center for Health Statistics, and other federal agencies are fundamental to meeting new funding requirements to share and de-identify data, and are often referred to in the literature on data sharing. We describe how these principles are employed by the Disclosure Review Boards (DRBs) of these two agencies, and then state these principles in more general terms that are applicable to any disclosure review process. The kinds of data that academic institutions share may call for less complex or stringent DRBs and specific nondisclosure procedures different from those employed by federal agencies, but the same general principles apply. Specific application of these six principles by non-government researchers will depend on the nature of their data, their own institutional resources, and the likely future usefulness of their data.

SDC in the 2000 U.S. Decennial Census

This paper describes the statistical disclosure limitation techniques to be used for all U.S. Census 2000 data products. It includes procedures for short form tables, long form tables, public use microdata files, and an online query system for tables. Procedures include data swapping, rounding, noise addition, collapsing categories, and applying thresholds. Procedures for the short and long form tables are improvements on what was used for the 1990 decennial census. Several procedures we will be using for the microdata are new and will result in less detail than was published from the 1990 decennial census. Because we did not previously have the online query system for tables, all of those procedures are newly developed.

The microdata analysis system at the U.S. census bureau

The U.S. Census Bureau collects its survey and census data under Title 13 of the U.S. code, which promises to protect the confidentiality of our respondents. The agency has the responsibility to release high quality data products without violating the confidentiality of our respondents. This paper discuses a Microdata Analysis System (MAS) that is currently under development at the Census Bureau. We begin by discussing the reason for developing a MAS, and answer some questions about the MAS. We next give a brief overview of the MAS and the confidentiality rules within the system. The rest of this paper gives an overview of the evaluation of the universe subsampling routine in the MAS known as the Drop Q Rule. We conclude with some remarks on future research.

Privacy and confidentiality resources.

Several organizations in the United States have a major interest in creating, testing, and using methods of data presentation that respect privacy and assure confidentiality. The following are among those that do so, and provide up-to-date information on these topics for the benefit of others who conduct human research: (1) The Committee on Privacy and Confidentiality of the American Statistical Association; (2) an interagency committee of the federal government, the Federal Committee on Statistical Methodology, and its subcommittees, the Confidentiality and Data Access Committee and the Committee on Privacy; (3) the Inter-university Consortium for Political and Social Research (University of Michigan), whose core mission is to archive important social science data, provide open and equitable access to data, and promote the effective use of data; and (4) Carnegie Mellon Universitys Department of Statistics, which has created an open-access online journal, the Journal on Privacy and Confidentiality. These resources are described, and URLs are provided to give readers web access to these resources.