Page 24 - Invited Paper Session (IPS) - Volume 1

IPS30 Diana B.
characteristics necessary for census. It has been stressed that the choice of the
register-based census methodology requires virtually all mandatory census
characteristics to be covered by registers, that there is a system whereby all
objects observed by census have been identified and that address data are
used.
The issue of registers emerged in the Statistics Estonia correspondence of
authorities again in 2005 – a year when systematic exercise of assessing the
quality of registers can be observed. The reason why the topic of registers was
not addressed could be associated, among other things, with strict legal
environment for producing statistics, by which e.g. it was prohibited by law to
use the 2000 census data for analysis (this applied until the entry into force of
the new act in 2010). This meant that data validation work using register data
was not possible either.
A new initiative was launched in 2007. This involved preparations for a new
census round. Statistics Estonia analysed the availability of the population and
housing census indicators in state databases on the basis of a self-assessment
questionnaire of the registers (11). The study concluded that the use of register
data, considering the compiled census programme, was not feasible due to
the fact that databases and registers did not have sufficient information for
the indicators required for the census (Paut, 2007). The main shortcomings
concerned the execution of the census programme, which consisted of the
mandatory output of the EU, as well as the needs of Estonian users. There was
a complete lack of data on household composition and actual family status,
working time, religion, foreign language skills and on the agricultural activities
of households. For the whole population, there was no data on occupation,
mother tongue, links between households and household members, the
number of children born to woman, educational attainment, migration and the
place of birth of parents. The data indicating the living conditions was lacking.
It was unknown whether the actual place of residence and the registered place
of residence of persons coincided. There was no overview of people living in
institutions (monasteries, children’s homes, etc.). It was not possible to link the
population register to the register of construction works and buildings; in
addition, a situation was possible that a person was registered in a dwelling
which did not exist; also, metadata in the registers were incomplete.
It was hard to explain to the public and the registrars that data needed for
the census were not available in the registers. It took some explaining before
it became clear that it was too early to have a register-based census.
In conclusion for the period 1996–2008, it should be noted that there were
two approaches to the use of registers for census: a pessimistic one,
emphasising a lack of data quality and big data gaps in the registers, and an
optimistic one, which highlighted the possibility to save on census costs by
using modern information technology tools and possibilities to integrate data

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