Finite Sample Inference for Multiply Imputed Synthetic Data Under a Multiple Linear Regression Model

• Activity Number: 591 - Synthetic Data and Data Disclosure
• Type: Contributed
• Date/Time: Wednesday, August 1, 2018 : 2:00 PM to 3:50 PM
• Sponsor: Government Statistics Section
• Author(s): Martin Klein*
• Companies: U.S. Census Bureau
• Keywords: Partially synthetic data; Pivotal quantity; Plug-in sampling; Posterior predictive sampling; Statistical disclosure control

Abstract

In this paper we develop finite sample inference based on multiply imputed synthetic data generated under the multiple linear regression model. We consider two methods of generating the synthetic data, namely, posterior predictive sampling and plug-in sampling. Simulation results are presented to confirm that the proposed methodology performs as the theory predicts, and to numerically compare the proposed methodology with the current state of the art procedures for analyzing multiply imputed partially synthetic data.