Page 316 - Special Topic Session (STS) - Volume 3

STS547 Owen A.
Figure 11: Percentage difference from census estimates for males by five-
year age group
England and Wales, 2011

Source: Office for National Statistics
Further work is being undertaken to refresh the simulation framework from
the earlier studies to enable a fuller evaluation of the performance of all of
these versions of capture-recapture estimators.

4. Discussion and Conclusion
Capture-recapture methods depend heavily on the underlying
assumptions being met, and when they are not estimates based upon them
are biased. This is clearly not desirable when using such techniques for official
statistics about the size of the population. This paper has outlined various ways
that National Statistical Institutes are attempting to tackle this issue in the
context of using administrative data to drive such estimates. Essentially, it is
an exploration of ways in which the assumptions can be balanced in different
ways according to the likely qualities of the data.
The methodological challenges of transforming population statistics
should not be underestimated. ONS is devoting significant resources to
exploring what might be possible, and this paper only touches the surface of
those that need and are being addressed. For instance, other methodological
challenges not covered here include data linkage, disclosure control,
uncertainty measurement and the output definitions. Research is into these is
continuing apace.

References
1. Abbott, O., Castaldo, A., Racinskij, V., Ross, H., Smith, P. A. and Brown, J. J.
(2015) Developing a weighting-class approach for the 2021 Census.
Accessed at:
www.researchgate.net/publication/305710321_Developing_a_weighting-
class approach for the 2021 Census

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