Introduction

Abstract

National statistical institutes (NSIs) such as the U.S. Census Bureau or the German Federal Statistical Office gather valuable information on many different aspects of society. Broad access to this information is desirable to stimulate research in official statistics. However, most data obtained by the institutes are collected under the pledge of privacy, and thus the natural interest in enabling as much research as possible with the collected data has to take a back seat to the confidentiality guaranteed to the survey respondent. But not only legal aspects are relevant when considering disseminating data to the public. Respondents who feel their privacy is at risk might be less willing to provide sensitive information, might give incorrect answers, or might even refuse to participate completely – with devastating consequences for the quality of the data collected (Lane, 2007). Traditionally, this meant that access to the data was strictly limited to researchers working for the NSI. With the increasing demand for access to the data on the micro-level from external researchers, accelerated by the improvements in computer technology, agencies started looking for possibilities to disseminate data that provide a high level of data quality while still guaranteeing confidentiality for the participating units.