Abstract:
|
Dissemination of public use microdata promotes informed policymaking and scientific studies. However, when the collected microdata is subject to regulatory privacy laws, it is often challenging to produce public use microdata that preserves complex inter-item relationships and protects the privacy of individual respondents, simultaneously. The proposed research aims to develop a statistically integrated microdata production system for selected economic programs conducted by the U.S. Census Bureau, focusing on the Economic Census (EC). The new production system will integrate editing, imputation, privacy protection (synthesis), and calibration stages using a single, coherent statistical framework. This approach preserves correlation structure, allowing for more sophisticated analyses than simple tabulations. In this talk, the application of the method to Economic Census and practical issues to make the synthetic data will be introduced.
|