Julia Simonson, Laura Romeu Gordo & Nadiya Kelle, Statistical Matching of the German Aging Survey and the Sample of Active Pension Accounts as a Source for Analyzing Life Courses and Old Age Incomes [Abstract]
The paper examines the combination of the German Aging Survey (Deutscher Alterssurvey – DEAS) with the Sample of Active Pension Accounts (Versicherungskontenstichprobe – VSKT), as an example of how survey results may be linked together with administrative data using statistical matching. Statistical matching is a technique increasingly being applied in order to combine information from different data sources where no linkage can be made between records based on any unique identifier. This might be due to confidentiality restrictions or attempts to avoid the high attrition rates connected with informed consent requirements. The aim of this matching is to provide a combined dataset that contains more information than the data sources would on their own. In our paper, we detail some preparatory steps for making this match, such as the definition and adjustment of matching variables. The steps we describe may well be worth challenging on the basis of the divergent characteristics of the two sets of data. We also outline the procedure we used to combine the data sources, based on the Mahalanobis distance vector. Finally, we assess the quality of the matching by comparing the individual pension amounts that we can extract from each of the two matching sources as our external criterion.