Census undercount and the Post Enumeration Survey
The Census undercount is the number of people the Census should have counted, less the number it did. The ABS measures it with the Post Enumeration Survey, run straight after each Census. In 2021 the net undercount was 0.7% of Australia's population, the lowest on record.
How the ABS measures it
The ABS has run a Post Enumeration Survey shortly after every Census since 1966 to assess independently how complete the count was. In 2021 the survey, also called the Post Census Review, selected about 45,000 households at random and collected information from more than 117,000 people. Interviewers asked basic demographic questions and where each person was on Census night, and the ABS matched those records to Census records to see whether, where and how many times each person had been counted.
Some people are missed (gross undercount) and some are counted twice or in error (gross overcount). The net undercount is the difference between the survey's estimate of how many people should have been counted and the Census count, including people the ABS imputed for dwellings that didn't return a form. In recent Censuses it has been positive, because more people were missed than overcounted.
The 2021 results
The 2021 Census counted 25,417,978 usual residents against a survey estimate of 25,608,022, a net undercount of 190,044 people or 0.7%. This was down from 1.0% in 2016, 1.7% in 2011 and 2.7% in 2006. By state and territory, the net undercount was:
- Northern Territory: 6.0%
- Western Australia: 3.0%
- Tasmania: 1.6%
- Queensland and South Australia: 1.0% each
- Victoria: 0.3%
- New South Wales: 0.0%
- ACT: a net overcount of 0.6%
Coverage differed sharply between groups. The net undercount of Aboriginal and Torres Strait Islander people was 17.4%, or 170,752 people, almost unchanged from 17.5% in 2016. Males were undercounted by 1.3% and females by 0.2%, and men aged 20 to 24 by 3.0%. Every age group under 40 had a net undercount above 1.0%, while every age group from 60 had a net overcount, reaching 2.0% for people aged 85 and over.
What it means for area figures
The Census figures published for suburbs, councils and statistical areas are not adjusted for undercount, though they include people the ABS imputes for dwellings that didn't return a form. The ABS applies the survey's results when it builds the estimated resident population, and it advises that its undercount factors should not be used alone to derive an alternative population figure, because ERP uses further data and adjustments.
For a site decision, undercount matters most where the groups it affects are concentrated. In an area with a large Aboriginal and Torres Strait Islander population, many young adults or a high share of people born in New Zealand (undercounted by 8.5%), the Census count is likely to sit further below the true population than in an older suburb, where the count can run slightly above it.
The 2026 Census was held on 11 August 2026, and the ABS planned its 2026 Post Census Review for immediately afterwards, with a sample of around 57,000 households and results due in June 2027.
Undercount and Censible
Censible's Census figures are the counts as the ABS published them, without adjustment for undercount. Its 2026 and 2031 population projections use the estimated resident population only as a growth rate from 2021 to mid-2025, so they stay on the Census's footing, and a change from 2021 reads as growth rather than as the gap between the Census count and the ERP.
Sources: ABS, 2021 Census overcount and undercount (released 28 June 2022); ABS, Census Post Enumeration Survey; ABS, Proposal for the 2026 Census of Population and Housing and 2026 Post Census Review.
Estimated resident population (ERP)
The ABS's official measure of Australia's population. It starts from the Census count, adds people the Census missed and residents who were overseas, and is updated between Censuses with births, deaths and migration.
Place of usual residence and place of enumeration
Two ways the ABS counts people in the Census. Place of usual residence counts people where they usually live, generally the address where they have lived or intend to live for six months or more. Place of enumeration counts them where they were on Census night.
Perturbation (small-cell adjustment)
The small random adjustments the ABS makes to Census counts so that no individual can be identified. The ABS calls it introduced random error. It barely affects large counts, though it can distort very small ones, and it means table totals may not equal the sum of their parts.
Carried figure
A count from an earlier Census moved onto newer boundaries with the ABS's population-weighted correspondences, so places can be compared across Censuses on the same map. Where a boundary didn't change, the carried figure is the ABS's own count; where it did, it's an estimate.
Questions people ask
The ABS's Post Enumeration Survey put the 2021 net undercount at 190,044 people, or 0.7% of the population, the lowest rate recorded for an Australian Census.
It is a large household survey the ABS runs straight after each Census to check how complete the count was. People's answers are matched to Census records to estimate how many were missed or counted more than once.
In 2021 the net undercount was 17.4% for Aboriginal and Torres Strait Islander people, 8.5% for people born in New Zealand, 6.0% in the Northern Territory and 3.0% for men aged 20 to 24, against 0.7% overall. People aged 60 and over had a net overcount.
Census counts for areas are published as counted, without adjustment. The ABS uses the survey results when it calculates the estimated resident population, which is its official population figure.
The ABS planned to release results from the 2026 Post Census Review in June 2027.