Conferences

The challenges of protecting subjects in public-use microdata while maximizing the analytic utility of these data have been discussed in the disclosure literature. This paper describes a new version of Survey Documentation and Analysis (SDA), an online data analysis program that performs bivariate cross-tabulation, comparison of means, correlation and regression analyses. Offering the same analytic capabilities as SDA, Secure SDA (SSDA) * is designed to provide a safe, reliable way to publicly distribute restricted-use data, thereby democratizing access to data previously unavailable or that required special procedures to obtain. SSDA’s development focused on three areas: enhancing analytic capability, programming options to restrict both access and analytic output and determining how to provide maximum electronic security of SSDA data. SSDA has improved its analytic capacity with the proper calculation of variances to adjust for complex sample designs in means and tables analyses. To provide maximum flexibility and utility, SSDA automates several disclosure protections that prevent the use of organization-defined high-risk variables, singly or in combination, as well as output (e.g. unweighted n’s). When the organization-defined rules are violated by an attempted analysis, the resulting output is fully or partially suppressed. The Web server structure was configured to provide optimal system security. These protections along with SSDA’s features and analytic capabilities are illustrated using data from the National Survey on Drug Use and Health (NSDUH). Future enhancements to the SSDA system are also discussed.