Effects of Changes in Data Collection Mode on Data Quality in Administrative Data. The Case of Participation in Programmes Offered by the German Employment Agency

"Der Einfluss des Datenerhebungsmodus auf die Datenqualitat von Verwaltungsdaten. Eine Methodenstudie am Beispiel deutscher Arbeitsmarktdaten". Until administrative data are available as research datasets, they are passed through many organizational units and stored in different formats. The transformations of collected data to a data warehouse and further the integration of data from several operational sources to an integrated dataset for research projects include various mappings of identifiers and variables. A particular challenge arises, whenever one of the intermediate products changes. The resulting difficulties are not only technical in nature, but may well lie in aspects of the theoretical interpretation in a particular research context. For long-term research projects, it is essential to ensure comparability between several versions of this dataset. So the main task resulting from changes in the data sources is to ensure that observations of a former version of a research dataset can be identified after these changes. A case study of the Integrated Employment Biographies (IEB) is presented as an example of these problems. In a first step reasons for changes in the data sources and the methodological problems of transformation between several versions of a research dataset are highlighted. In a second step some tests of variables fundamental for research analysis and stratification are presented.