How to use data swapping to create useful dummy data for panel datasets

"Many research data centres (RDCs) provide access to micro data by means of onsite use and remote execution of programs. An efficient usage of these modes of data access requires the researchers to have dummy data, which allows them to familiarize with the real data. These dummy data must be anonymous and look the same as the original data, but they do not have to render valid results. For complex datasets such as panel data or linked data, the creation of useful dummy data is not trivial. In this paper we suggest to use data swapping with constraints in order to keep some consistency and correlation between variables within crosssections and over time. It is easy to be implemented even for datasets with many variables and many survey waves." (Author's abstract, IAB-Doku) ((en))