Page 12 - Invited Paper Session (IPS) - Volume 1

IPS30 Stefan D. et al.

The use of registers in the German censuses
2011, 2021 and beyond
Stefan Dittrich, Thomas Koerner, Markus Zwick
Federal Statistical Office, Germany

Abstract
The history of population and housing censuses in Germany is special for a
number of reasons. The most obvious feature is that after the last traditional
census in 1987 the next population census did only take place in 2011. The
census in 2011 was prepared under the restrictions of a judgment of the
German constitutional court that preceded the census in 1987 and imposed
strict data protection rules. Still under the impression of the public debate in
1987, in 2011 a combined census model was launched integrating elements
of register use, a conventional census (on housing), and a sample survey. The
results from the sample survey are used in order to correct the data for errors
due to over- and undercoverage as well as to collect such variables that were
not available from registers. The paper, in a first part, gives insight into the
backgrounds to the creation of this specific model and provides an overview
of the experiences made in the 2011 census. In its second part, the paper
presents the lessons learned for the implementation of a similar combined
model in 2021 and introduces the current plans to move to a purely register-
based census model in the post-2021 era.

Keywords
Combined census; register-assisted census; register-based census; correction
of over- and undercoverage in population registers; census post-2021
developments

1. Introduction
Introducing a combined census model in Germany was not straightforward
for several reasons. Firstly, fewer suitable registers are available than in
countries that are adopting register-based approaches. Secondly, strict data
protection regulations that were established in the context of the last
traditional German census in 1987 make it challenging to find solutions to link
registers from different areas. Against this background neither a person ID nor
a dwelling ID have been introduced so far, which makes any linkage between
registers a burdensome undertaking. At the same time, a traditional census
based on interviewer-administered data collection is not popular among
stakeholders due to the sheer size of the cost.
Since 1983, the traditional census data collection has also enjoyed only
limited popularity among the respondents. The last traditional census in

1 | I S I W S C 2 0 1 9