2021 Census overcount and undercount

Census night was Tuesday 10 August 2021. All people in Australia should have been included on a Census form at the place where they stayed on this night (except foreign diplomats and their families). Shortly after the Census, the ABS ran a Post Enumeration Survey (PES) to independently measure how well this was achieved.

The ABS used PES results to determine how many people should have been counted in the Census (PES population estimate), how many people were missed (undercount) and how many people were counted more than once or in error (overcount). We combined these measures to produce the net undercount of the Census.

A new measure has been introduced showing the difference between the PES population estimate and the number of unique persons from which Census received a response. This is known as gross coverage error.

All rates presented in this release are a proportion of the PES population estimate of a given population.

The 2021 Census counted 25,417,978 Australian residents who were in Australia on Census night (including people imputed for non-responding dwellings).  The PES estimate for the same population was 25,608,022 persons. This represents a net undercount rate of 0.7%, the lowest recorded net undercount rate for an Australian census.

Note: Census counts in this release correspond to the scope of the PES.

The Northern Territory recorded the highest net undercount rate of all states and territories at 6.0%. The Australian Capital Territory recorded an overcount (a net undercount of -0.6%).

Net undercount decreased between 2016 and 2021 in New South Wales, Victoria and Queensland. These jurisdictions were subject to COVID-19 related restrictions at the time of the Census and PES, limiting peoples’ movement. In each:

• There was a decrease in gross undercount (fewer people missed) and reduced over-imputation (net overcount in non-contact sector). Together these drove the overall decrease in net undercount.
• Gross coverage error was lower, representing higher Census coverage.

Compared with 2016, net undercount increased in South Australia, Western Australia and Tasmania in 2021. In all these states there was a decrease in net overcount in the non-contact sector, reflecting lower levels of imputation compared with 2016.

• In South Australia, the net undercount rate increased from 0.2% in 2016 to 1.0% in 2021, while gross coverage error remained stable.
• In Western Australia, the net undercount rate increased from 0.4% in 2016 to 3.0% in 2021, and gross coverage error also increased.
• In Tasmania, the net undercount rate increased from 0.1% in 2016 to 1.6% in 2021, with a small increase in gross coverage error.

In the Northern Territory, the net undercount increased from 5.0% in 2016 to 6.0% in 2021. The gross coverage error of 14.3% was lower than 2016 meaning Census received a response from a higher proportion of the population in 2021.

Similar to 2016, the Australian Capital Territory recorded a small net overcount (-0.6%) in 2021. As with other eastern seaboard states, this jurisdiction was subject to COVID-19 related restrictions at the time of the Census and PES, limiting peoples’ movement. Gross coverage error remains the lowest gross coverage error of all the states and territories.

Nationally, the net undercount rate for capital cities was lower (0.5%) than that for rest of state regions (1.2%). This was a reversal of the net undercount rates in 2016 (1.2% net undercount in capital cities and 0.5% in rest of state regions) and was driven by decreases in the net undercount rate for Sydney, Melbourne, Brisbane and Canberra. Gross coverage error in these capital cities also showed decreases from 2016 indicating higher Census coverage in these areas.

In Adelaide, the result changed from a small net overcount in 2016 to a small net undercount in 2021, while the net undercount decreased for the rest of South Australia. Gross coverage error remained stable across both regions.

The net undercount rate and gross coverage error increased in both capital cities and rest of state regions for Western Australia and Tasmania.

While the net undercount rate was slightly higher in Darwin compared to 2016, the gross coverage error was more than two percentage points lower. Both net undercount and gross coverage error increased in the rest of the Northern Territory.

Young adults are more likely to be missed in a census, while older adults are more likely to be counted or overcounted, as observed by previous censuses (both in Australia and overseas). Young children are often mistakenly omitted from census forms, and males are traditionally more likely to be missed than females. This pattern held in 2021:

• More males were missed by the Census than females - the net undercount rate for males was 1.3% compared with 0.2% for females. 
• Males aged 30-34 years had the highest net undercount rate (3.6%) of all age groups.
• Females aged 85 years and over had the highest net overcount rate (-3.0%) of all age groups.
• All age groups under 40 years had a net undercount rate over 1.0%, with the highest net undercount rate for the 10-14 year age group (2.3%).
• All age groups 60 years and over had a net overcount with the highest net overcount rate in the 85 years and over group (-2.0%).

The 2021 Post Enumeration Survey estimated that 983,257 Aboriginal and Torres Strait Islander people should have been counted in the Census, compared with 812,505 persons who were counted. This is equivalent to a net undercount of 170,752 persons, or a rate of 17.4%.

While the net undercount in 2021 (17.4%) was similar to 2016 (17.5%), each of the components reduced.

• Gross undercount in the contact sector was lower than in 2016 meaning fewer Aboriginal and Torres Strait Islander people were missed by the Census (11.0% in 2021 compared with 11.6% in 2016).
• Gross overcount rate also reduced, indicating fewer people were counted more than once or in error (1.5% in 2021 compared with 2.1% in 2016).
• While still an overcount, the net difference in classification between PES and Census was smaller (-1.3% in 2021 compared with -2.5% in 2016). This is the overall effect of people having a different Indigenous status recorded in Census compared with PES.
• The undercount for Aboriginal and Torres Strait Islander people due to Indigenous status being not stated on returned Census forms was stable (0.7% in 2021 compared with 0.8% in 2016).
• The non-contact sector net undercount for Aboriginal and Torres Strait Islander people was lower (8.4% in 2021 compared with 9.7% in 2016).

Gross coverage error for Aboriginal and Torres Strait Islander people decreased from 2016 (19.6%) to 2021 (18.8%), reflecting improvements in the components of net undercount.

For further information on how to interpret estimates of net undercount for the Aboriginal and Torres Strait Islander population, see Methodology.

Net undercount rates are shown for the ten most commonly reported countries of birth for people residing in Australia, according to the 2021 Census.

• People born in New Zealand had the highest net undercount rate (8.5%), and people born in Italy had the lowest net undercount rate (1.8%).
• Gross coverage error for people born in New Zealand and South Africa was higher in 2021 compared with 2016. For people born in all other countries the gross coverage error was lower.

This data item was not imputed for non-responding (imputed) persons. For information on how to interpret estimates of net undercount by Country of birth, see Methodology.

Net undercount is important for an effective understanding of the completeness of Census counts. The undercount adjustment factor is an additional measure that provides an indication of how much the Census count for a given category would need to be adjusted in order to reflect the PES population estimate for that category.

The undercount adjustment factors should not be used alone to derive an alternative measure of the Estimated Resident Population (ERP), as the latter includes additional data and adjustments for usual residents of Australia.

06/02/2023 - In the Undercount adjustment factor section, the PES population estimate for ACT in the table has been updated from 421,707 to 451,707. This now reflects the correct numbers, as per the Excel data cube in the Data downloads section.