An Investigation of Model-Based Microdata Masking for Magnitude Tabular Data Release

Traditionally, magnitude tabular data and microdata masking have been treated as two independent problems. An increasing number of government agencies are exploring establishing remote data access centers where both types of data release may occur. We argue that in these cases, consistency across both types of data release becomes an important component in the assessment of the performance of a certain masking and a common approach to the problem of masking both tabular and microdata would produce better results than approaches that address the two problems separately. Along this line, in this study we investigate the efficacy of using a model based microdata masking method (specifically Data shuffling) when the data is also used for magnitude tabular data release. We identify some aspects of our proposal that are important in addressing this issue further to perform a comprehensive evaluation of techniques suitable for both microdata and magnitude tabular data release.