Effects on Analysis Arising from Confidentialising Data Using Random Rounding

Abstract

Government statistical agencies collect data on individuals. These data can have personal information that will lead to individual identification. The information gathered is often released and used by other agencies. In order to preserve confidentiality (people’s privacy) the data are treated in a way that prevents identification. In recent years there has been a rapid increase in the research in the area of confidentiality and statistical disclosure techniques (SDC). We focus on random rounding method, one of the SDC. In this thesis we use rounded data which have been collected by Statistics NZ. We examine the effect of random rounding in contingency tables. We simulate data, based on rounded data, and actual data and use the general log-linear model and chi-square test for analysis.