FDZ-Arbeitspapier Nr. 7

A standard for the release of microdata

Statistical Offices in Germany may provide microdata to the scientific community, if these data are sufficiently anonymized. We present a standard for evaluating the degree of protection of a confidential data file. In a first step distance based record linkage is used to re-identify statistical units of the confidential target data. Besides re-identification of the unit it is also important to look at the benefit to a potential data intruder in case he reveals information. The more the information in the data disseminated is disturbed the lower is the benefit a data intruder derives from re-identification. For this reason, in a second step the re-identified units are analyzed if they contribute benefit to potential data intruders.

The paper shows how the standard mentioned can be applied to real world examples, taking the German Turnover Tax Statistics (almost full survey, about 3 million units), the German Structure of Costs Survey (a sample containing about 18 000 units) and the German Retail Trade Statistics (a sample containing about 23500 units) as a basis. Recently, so called Scientific-Use-Files of these surveys have been made available for the scientific community.