A Smoothing Approach to Data Masking

• Activity Number: 192
• Type: Contributed
• Date/Time: Monday, August 4, 2008 : 2:00 PM to 3:50 PM
• Sponsor: Section on Government Statistics
• Author(s): Yijie Zhou*+ and Francesca Dominici and Thomas A. Louis
• Companies: Merck & Co., Inc. and Johns Hopkins University and Johns Hopkins University
• Keywords: Data Masking ; Confidentiality ; Spatial Smoothing

Abstract

Individual-level data are often not publicly available due to confidentiality. Instead, masked data are released for public use. However, analyses performed using masked data may produce biased parameter estimates. We propose a data masking method using spatial smoothing. The method allows for varying both the form and the degree of masking by utilizing a smoothing weight function and a smoothness parameter. We investigate for GLM the bias of parameter estimates resulting from analyses using the masked data, and we show that data masking using a smoothing weight function that accounts for prior knowledge on the spatial pattern of exposure may lead to less biased estimates. We apply the method to the study of racial disparities in mortality, and we find that the bias of the association estimate when using the masked data is highly sensitive to both the form and the degree of masking.