September 18, 2014
Stefan Bender (IAB)
It is well known, that each data source has to deal with its potential sources of errors. Surveys have to deal with the challenges of non-response or memory errors of the persons interviewed, just to mention two. Survey methodologists have offered an advanced approach to deal with these challenges: the total survey error framework. In contrast, some variables in administrative data have a high accuracy. Due to the consistent process of data generation over time past information is as accurate as actual information. Although the image of administrative data is very positive (“gold standard”) not really much is known about the process of data generation. Most of the administrative data is not collected for research purposes, thus the definitions of variables are not always fitting into the researchers’ needs or fundamental information is missing. Linking survey with administrative data is one of the promising ways to take out the best of both worlds. If administrative data – for example – are used as the sampling frame for surveys, a lot can be known about the non-participants of surveys. Because of data protection reasons in Germany, obtaining an informed consent from the interviewee for the linkage is mandatory. Of course, as not everyone agrees to the linkage selection process may occur. The presentation will show some examples of linking administrative data of the Institute for Employment Research (IAB) of the Federal Employment Agency (BA) in detail with respect to the linkage process and possible selections.. At the end of the presentation the question will be raised, if informed consent is – in the light of big data – still the appropriate approach for protecting privacy.
Presentation (1.73 MB)
About the speaker
Stefan Bender is head of the Research Data Center (FDZ) of the German Federal Employment Agency (BA) in the Institute of Employment Research (IAB) and vice-chair of the German Data Forum (Rat für Sozial- und Wirtschaftsdaten). He is member of the standing committee in the EU 7th Framework Programme project “DWB: Data without Boundaries”. His research interests are data access, data quality, merging administrative, survey data and/or big data, record linkage, unemployment, minimum wage, management quality, quality of work and job loss. He has published in leading journals like the American Economic Review, Quarterly Journal of Economics, European Economic Review or International Journal of Epidemiology. He is co-editor of the book “Privacy, Big Data, and the Public Good: Frameworks for Engagement”, Cambridge University Press (together with Lane J., Stodden, V. and Nissenbaum, H.), 2014.