Likelihood-Based Finite Sample Inference Based on Synthetic Data

• Activity Number: 138
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Survey Research Methods Section
• Author(s): Bimal Sinha*+
• Companies: UMBC
• Keywords: synthetic data ; likelihood based analysis ; noise multiplication

Abstract

When statistical agencies release micro data to the public, a major concern is the control of disclosure risk while ensuring utility in the released data. We consider likelihood based finite sample inference based on synthetic data for three probability models: exponential with an unknown mean, and normal with an unknown mean and either known or unknown variance, leading to some new results and conclusions! A comparison of this analysis with the one under noise multiplication also reveals some interesting features.