In view of declining response rates when collecting survey data, researchers are faced with the challenge of designing surveys in such a way that not only enough people participate, but also that these interviewees are representative of the underlying population. Due to the high complexity of the survey questions and the required sample sizes, most surveys are not only very costly, but also take a lot of time to plan, test, and conduct. To reduce these efforts, but also to minimize the burden on respondents, attempts are increasingly being made to draw on additional data sources.
Administrative data has proven to be particularly useful, as it usually contains large amounts of valid and reliable data. A disadvantage of administrative data, however, is that it only contains information required to fulfil its original administrative purpose. Therefore, this kind of process data can often not be used alone for answering research questions, which is why the linking of survey and process data (record linkage) has become increasingly common in recent years. In this way, important contextual information, which is usually missing in the process data, can be collected, and supplemented as part of a survey.
Although record linkage offers many advantages for researchers, merging different data sources also requires an awareness that the linkage includes selection processes that can have a negative impact on data quality. As a record linkage of survey and process data from the German Pension Insurance has been carried out as part of the study on “Life courses and old-age provision” (Lebensverläufe und Altersvorsorge”; LeA), various projects investigated these selection processes. For example, the extent to which record linkage consent varies with socio-economic determinants is examined. Furthermore, the different stages of the record linkage process at which possible sources of error can occur and how these errors can be minimized are analysed. The main aim here is to compare different record linkage projects and to identify best practice solutions for minimizing possible errors and resulting biases.
Kontakt
