Protecting Numerical Confidential Data using Data Shuffling: A Demonstration of Effectiveness of Approach and Flexibility of Delivery

Protecting numerical confidential data in data released by government agencies and other organizations poses a considerable challenge. There have been many different procedures that have been developed for this purpose. Among these procedures, Data shuffling offers the best compromise between preventing disclosure of information while preserving the usefulness of the data. Specifically Data shuffling offers the following advantages: (1) Individual data values are not modified but are shuffled between records thereby maintaining the distribution of individual variables exactly, (2) All linear and monotonic non-linear relationships between all variables are maintained providing a high level of analytical value, and (3) Providing the highest possible level of security (lowest level of both identity and value disclosure risk). In this demonstration, we provide an extensive analysis of the analytical value and disclosure risk characteristics of Data shuffling. We also demonstrate three modes of delivery based on the specific needs of the organization: (1) Excel based solution for small applications; (2) Web based solution for larger applications where the organizations wish to perform Data shuffling by themselves, and (3) Third party solution for organizations and agencies for large complex data sets. We believe that this demonstration will be useful for any organization that intends to analyze, share, or disseminate numerical confidential data without risk of disclosure.