Page 62 - Special Topic Session (STS) - Volume 4

STS563 Pete Jones
transformation at ONS. Since 2011, ONS has been working towards the UK
government ambition to deliver future censuses using administrative data. An
increasing number of countries are now aiming to transition from traditional
census collections in favour of reusing existing data collected by government.
For countries that have a population register, the existing pre-conditions are
in place to produce reliable population estimates. For the UK and other
countries that do not have a population register, there are considerable
challenges in compiling administrative data to have accurate coverage of the
resident population. The ONS have developed methods to link multiple
administrative sources, including health, tax and benefits, and education to
5
construct a Statistical Population Dataset (SPD) . Figure 2 below shows the
coverage patterns of the SPD population estimates relative to the 2011
Census.

Figure 2: Statistical Population Dataset (SPD) estimates compared to 2011 Census, England &
Wales

The comparison in figure 2 shows that the SPD is characterised by under-
coverage (typically a consequence of non-registration on administrative data),
but also over-coverage (for example, individuals that are registered on
administrative data but have since left the country). To measure and adjust for
these coverage biases, an ongoing Population Coverage Survey (PCS) is
proposed for collection on a continuous basis. The PCS is based on similar
6
concepts as the Census Coverage Survey (CCS) which has been used as a
post-enumeration survey to adjust for non-response bias in the 2001 and 2011
Censuses. An important difference regarding the CCS is that it has been
designed as an independent survey that is collected during the weeks
immediately following a Census collection. The aim of the PCS is to integrate
with wave 1 of the new Labour Market Survey, which will necessitate
continuous collection and redesign of the sampling approach.

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