Data Linkage in the Social Sciences - Hurdles and Overcoming Them

Data linkage between two different microdata sources uncovers huge potentials for research. However, enabling the linkage of confidential datasets between different institutions, authorities or research data centers proves to be difficult at best. Data linkage is limited by legal frameworks and data protection concerns or made impossible by a lack of a legal basis. Even if a research clause or informed consent to linkage exists the fear of re-identification of individuals in the data or the concern for explicit legal permission outweighs this in case of doubt. As a result, the potential of existing data sets cannot be optimally exploited for research in the social sciences, with consequences for the study of society, policymaking and Germany as a center of science in general. Despite these hurdles, linking data provides many benefits such as reducing survey time and frequency, thereby putting less burden on the respondent and reducing the costs of surveys. Moreover, the quality of surveys can be examined and enhanced by linking register data for plausibility checks. Finally, linked data entail more information in general, e.g. by bringing to-gether different subject areas like work and household decisions, corporate structures, trade and finances, or by adding longitudinal information, for example employment, migration or establishment histories, thus offering a rich data base for empirical analysis and scientific knowledge gain, but also for evidence-based policymaking. The importance of facilitating data linkage was recognized by the German government, which resulted in the proposed German Research Data Act [1]. The Act did not make it past the first legislative steps, but despite its flaws, it would have provided a legal basis for any register data collected by the Federal State in Germany. Still lacking a legal basis for many linkage projects, we outline the hurdles that have to be overcome and draw on positive examples to show how data linkage of sensitive data can be possible under the current legal framework. Thereby we highlight which preconditions must be met to make linkage possible. Since 2024, data from the Deutsche Bundesbank can be used for the first time with data from the Federal Statistical Office of Germany as integrated data sets by the scientific community via the RDC-StBA. All legal, technical and methodological challenges were overcome in a joint project between the two institutions and the scientific community. As another example, the RDC-IAB now hosts a data product linking survey data of the German Socio-Economic Panel (SOEP) with administrative data of the IAB, thus combining rich socio-demographic survey information with precise longitudinal data on employment and labor income. However, these examples show that despite the legal permissibility of linking the data, the practical im-plementation is very time-consuming and complex.