With some controversy surrounding undercount recently, it is a good time to explain the issues surrounding underenumeration or undercount
from the 2001 Census.
While the census makes every effort to count each person once, some people are missed and some are counted multiple times. Generally more people are missed than are counted more than once, and this leads to a net undercount.
People may have not been counted because the dwelling they were in on Census night was missed because it was difficult to find. Others may have been missed, even though the dwelling they were in was counted, due to insufficient space on the household form or privacy concerns.
People may have been counted more than once due to being included on the census form at the dwelling where they usually lived even though they were also counted elsewhere on Census night.
A measure of the extent of net undercount is obtained from the Post Enumeration Survey (PES), conducted soon after the census. The PES revealed that the 2001 Census missed 1.8% of people in Australia on Census night. However, rates of undercount are not uniform and can vary significantly according to age, sex, ethnicity and geographic area.
An allowance for net undercount based on the results of the PES is added to the census count when calculating the Estimated Resident Population (ERP), which is used for allocating seats in parliament and distributing billions in Government funding. However, the census counts themselves are not adjusted.
The results of the PES also provide an assessment of census procedures and allow improvements to be made for the next census.
More information about undercount in the 2001 Census can be found in the Information Paper: Data Quality - Undercount (cat. no. 2940.0), available free of charge on the Census webpages.
Further information about undercount can also be found in:
- Demography Working Paper 2003/2 - 2001 Net Undercount
- Demography Working Paper 2002/2 - Estimated Resident Population and Effects of Census Systems Created Records
- Fact Sheet: Effects of Census Processes on Non-response Rates and Person Counts