Generating multiply imputed synthetic datasets : theory and implementation. Gerhard-Fürst-Award

Abstract

"Over the last few years, various different variants have been suggested in the relevant literature for the creation of synthetic data. Within the framework of the dissertation presented here, these processes were compared to one another and each in turn applied to the Establishment Panel of the Institute for Employment Research (IAB) of the Federal Employment Agency. An important result of this work is the synthetic data sets of the wave 2007 of the IAB Establishment Panel that have been available since the beginning of 2011 via the Research Data Centre (FDZ) of the Federal Employment Agency at the Institute for Employment Research (IAB). In addition to this, a new, two-step imputation procedure is presented which allows a better weighing up between the re-identification risk and an as-high-as-possible quality of data. Along with this, new measures are proposed in order to measure the remaining re-identification risk of the synthetic data sets. The aim of the article is to give a brief presentation of the individual variants of the procedure along with important findings of the research." (text excerpt, IAB-Doku) ((en))

Cite article

Drechsler, J. (2011): Erzeugung synthetischer Datensätze durch multiple Imputation. Theorie und Implementierung in der Praxis. Gerhard-Fürst-Preis. In: Wirtschaft und Statistik No. 4, p. 402-407.