Associate Director for Strategic Planning and Innovation U.S. Census Bureau

Traditionally, national statistical offices (NSOs) have released tabulations, and some have released public-use microdata files that have been protected against disclosing confidential information by using techniques like top-coding variables and limiting geographic detail. More recent methods of disclosure avoidance include creating synthetic data and noise addition. Because not all research and information needs can be met by these tabulations and data files, NSOs have developed three key alternatives for providing non-government researchers with access to confidential microdata to improve statistical modeling. The first, licensing, allows qualified researchers access to confidential microdata at their own facilities, provided certain security requirements are met. The second, statistical data enclaves, offers qualified researchers restricted access to confidential economic and demographic microdata at specific agency-controlled locations. Thirdly, statistical agencies can offer remote access, through a computer interface, to the confidential data under automated or manual controls so that only disclosable results of the analysis are returned to the researcher, who is unable to view or copy any of the confidential data