2940.0.55.002 - Information Paper: Measuring Overcount and Undercount in the 2016 Population Census, Jul 2016  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 01/07/2016  First Issue
   Page tools: Print Print Page Print all pages in this productPrint All



In Australia, estimates of overcount and undercount in the Census are based on the results of the Census Post Enumeration Survey (PES). The purpose of the PES is to determine how many people were counted more than once in the Census (and how many times these people were counted) and how many were missed.

The PES is a household survey conducted by specially trained interviewers starting about six weeks after Census night. This is a different collection methodology to the Census, where most forms are completed by any responsible adult on behalf of the household. A major advantage of interviewer-administered questionnaires is that people can be provided with assistance if they are uncertain about the meaning of any questions.

The scope of the PES includes all private dwellings in all areas, including Discrete Communities, but excluding special dwellings. For each dwelling selected in the PES, the any responsible adult (ARA) methodology is used as it is in Census. This generally involves a single person in the dwelling responding on behalf of all persons present or usually resident. In addition to obtaining basic demographic information, questions are asked about each person's usual residence, location on Census night, and any other addresses where they might have been counted in the Census.

Using this address information, the corresponding Census forms are examined to confirm how many times each respondent in the PES was counted in the Census. The results of this process are then combined and weighted to produce an estimate of net overcount or net undercount in the Census.

In the 2011 PES a sample of 36,000 private dwellings responded to the PES, about half of 1% of all private dwellings in Australia. The total number of people included in the PES was 94,000. The sampling fraction varied between states and territories, with larger sampling fractions in the states and territories with smaller populations. This design provides reliable estimates for each state and territory, as well as for Australia as a whole. Sampling fractions are also larger in Discrete Communities compared to the rest of the state they are in to ensure adequate reliability of Aboriginal and Torres Strait Islander estimates.

The 2016 PES sample size is larger than that for 2011, with a target of 43,000 responses from private dwellings. Additional information on the 2016 sample is provided in the Sample Design section of this paper.