5. Conclusions and observations

The 2021 Census Statistical Independent Assurance Panel (the Panel) has concluded that the 2021 Census data is fit-for-purpose, is of comparable quality to the 2011 and 2016 Censuses and can be used with confidence.

In assessing whether the 2021 Census was fit-for-purpose, the Panel considered how the data compared with previous Australian and international censuses. Internationally, the 2021 Australian Census was comparable to those of Canada and the UK, and higher than the rate for New Zealand which was conducted pre-pandemic. The high response rate for private dwellings (96.1%) was an outstanding achievement given the challenges provided by the pandemic. While the response rate was higher than that achieved in the 2016 Census, the Panel notes that some remote areas’ response rates have declined.

While some quality challenges exist in the 2021 Census, such as the undercount of Aboriginal and Torres Strait Islander peoples, these issues were also present in previous Australian censuses and have not changed significantly. It is also noteworthy that the 2021 Census faced the additional challenge of being conducted during the COVID-19 pandemic and managed to overcome this safely and without affecting data quality. The similarity of the 2021 Census data quality to that of the 2011 Census, and the observed improvement compared with the 2016 Census, contributes to the Panel’s assessment that the 2021 Census data quality is comparable and fit-for-purpose.

Nearly all people in Australia completed Census forms and this was a better result than observed in the 2016 Census. There was a significant increase in the use of the online Census form (with 78.9% of dwellings responding online) and this saw an associated increase in data quality, most noticeably by reductions in data item non-response rates. 

While the response rate for non-private dwellings is similar to 2016, adjustments to the framework used for non-private dwellings led to improvements in quality. There have also been improvements to the non-response imputation methods for non-private dwellings which have resulted in improvements in quality.

The 2021 Census data aligns well with expectations. Counts of the population at the national and state and/or territory level derived from the 2021 Census, adjusted by the Post Enumeration Survey, compare well to the Estimated Resident Population, Australia’s official population estimate, and can be used to rebase these estimates. During comparison of the Census and the unrebased Estimated Resident Population, the Panel noted the importance of five yearly censuses for recalibrating and rebasing the Estimated Resident Population to maintain its accurate reflection of population movement at the state and territory level and it is likely to be even more important at smaller geographic levels.[1] The Panel considers that the Census is clearly fit-for-purpose for all of its important uses and, in particular, the rebasing of the population estimates which are used to allocate funds to the states and territories and allocate numbers of seats for the House of Representatives between the states and territories. Other important uses of Census data include supporting the planning and administration of government, business and other users, and the provision of comprehensive information for small geographic areas and for population groups across Australia.

The 2021 Census continued to collect a comprehensive, detailed picture of Australia’s population. The Panel examined a number of key topics including population counts, Sex, Age, Income, counts of Aboriginal and Torres Strait Islander peoples, Country of birth, Language, Ancestry and Religion. The Panel’s analysis revealed that the items reviewed generally have results which are consistent with previous censuses or independent data sources or have observations consistent with expected societal change or pandemic impacts. Greater levels of response to most questions and an increased online response have resulted in improved data quality. The Panel consider Census topics examined fit-for-purpose but noted that further ABS quality assessments will be very important for interpreting some data items such as Long-term health conditions and Income.

Two new topics were added to the 2021 Census: Long-term health conditions and Australian Defence Force service. Their results generally aligned with the expectations from independent data sources, although the Panel observed some notable differences for further investigation and explanation.

Key quality indicators from the Post Enumeration Survey support the comparability of the 2021 and 2016 Censuses with some improvements observed. The net undercount rate for persons on Census forms decreased from 3.7% to 2.8% between 2016 and 2021, and their gross overcount has remained stable at 1.2%. There have also been improvements in other quality indicators from the Post Enumeration Survey such as the total net undercount. The Panel also noted that the Post Enumeration Survey was very successful given the need to significantly change the methodology because of COVID-19 restrictions on field visits.

The Panel concluded that the Census population data, adjusted by the Post Enumeration Survey, is fit-for-purpose for its important use in rebasing the Estimated Resident Population at the national and state and/or territory level. The Panel considers a five yearly census highly important to calibrating and rebasing Australia’s Estimated Resident Population, one of the most important uses of census data.

Notwithstanding increased efforts and investments by the ABS, one disappointment has been that the net undercount for Aboriginal and Torres Strait Islander peoples remains unchanged at 17% as estimated by the Post Enumeration Survey.

The Panel noted that there had been public concerns raised about privacy prior to the 2016 Census, and that one of their observations related to improved management of privacy. The Panel was pleased to observe that the ABS’s proactive and thorough approach to privacy for the 2021 Census saw few privacy concerns raised. The Panel concluded that improved response to the question on National Archives retention and increased agreement to retention suggests that the public approached the issue of privacy in the 2021 Census more positively than the previous Census.

The Panel notes that there are several areas that deserve specific consideration, as follows.

[1] The Panel is not in a position to make a judgment about data quality below the state and/or territory level due to time constraints.

The Panel considers that the ABS’s approach to conducting the 2021 Census during the pandemic was impressive and well managed. The 2021 Census was conducted at a time when many parts of Australia were in lockdown and movements within, into and out of Australia were tightly controlled. Through appropriate planning, risk management, and rigorous testing, disruptions to Census field operations due to the pandemic were largely overcome without impacting the quality of Census data or safety of the public and Census staff. Nevertheless, the ABS faced difficulties with recruitment reflecting a tight labour market, and other challenges caused by the pandemic, such as restrictions on the movement of field staff both intra and interstate.

New online self-service options, developed to improve the Census experience in 2021, were popular with households across Australia and proved particularly beneficial in helping minimise the impact of COVID-19 restrictions on the delivery of Census forms and other field work. The Census Digital Service was completely redeveloped for 2021 and its ease of use proved especially useful given the disruptions to field operations because of COVID-19 restrictions – people could easily access help and the online form, which they did, and in much larger numbers than in previous censuses.

The Panel observed the expected impact of COVID-19 on 2021 Census data with its impact on peoples’ movements within, into and out of Australia clearly reflected in 2021 Census data across a range of items, including the expected declines in foreign student populations and the impact of no immigration for an extended period. In addition, more people were counted at home in the 2021 Census than previously, due to reduced population movement prior to and during the time of the Census. This has enhanced data quality.

The Address Register is an increasingly important resource. It provided the frame for the 2016 and 2021 Censuses and, in 2021 served as the frame for the Post Enumeration Survey for the first time. The Address Register enables the use of an increasingly efficient and modern Census enumeration model with the delivery of login information directly to peoples’ homes. The Panel has observed however, that there are missing dwellings on the Census frame. Although the number of missing dwellings on the frame is rather small in relative terms, the reasons are worth investigating given the importance of the Register to a range of ABS collections.

The Panel observed that the improvement in response rates in 2021 was not uniform. While there were significant improvements in capital cities, especially the inner-city areas of most of our cities, there was some decline in response across regional and remote parts of Australia. This is likely to have been associated with recruitment difficulties due to tight labour markets, meaning that fewer, local field staff were on hand to assist in regional and remote areas, a difficulty that was exacerbated by COVID-19 restrictions which limited the movement of field staff around the country to fill in recruitment gaps.

The growth in the counts of Aboriginal and Torres Islander peoples seen in previous censuses has continued. Changes in the counts have an impact on other statistics, such as population measures and some performance indicators used for government reporting. As has occurred after previous censuses, ABS will further analyse the increase in the counts to understand the changes observed in the 2021 Census.

A large increase in online form response by Aboriginal and Torres Strait Islander peoples is expected to have associated increases in data quality consistent with the reasons mentioned elsewhere (see Section 3.3.4). A change to the layout of the Ancestry question for 2021 has resulted in increased numbers of people reporting Aboriginal and Torres Strait Islander ancestry.

Aboriginal and Torres Strait Islander peoples continue to be under-represented in the Census. The Post Enumeration Survey estimated the net undercount rate to be about 17% for each of the last three Censuses despite increased investment in the enumeration of this population. COVID-19 lockdowns and border closures had an impact on the availability of field staff within states and interstate and it is likely that this had an impact on enumeration of Aboriginal and Torres Strait Islander peoples in some areas. The improved enumeration of Aboriginal and Torres Strait Islander peoples should continue to be a priority area for attention by the ABS.

Non-response in the 2021 Census was lower than for 2016. The main way of compensating for non-response is by imputing for missing dwellings and persons. This requires identification of non-responding dwellings that were occupied on Census night as well as accurately imputing persons. If this could be done perfectly, then non-response would not be an issue. In practice, imputation cannot be done perfectly.

Occupancy determination was more difficult in the 2021 Census because of the inability to have personal contact with many households because of the pandemic. Notwithstanding this the Panel have observed from preliminary analysis of Post Enumeration Survey data that improvements have been made in occupancy determination and imputation between the 2016 and 2021 Censuses. Nevertheless, overcount in the non-contact sector (imputed records) is still driven by error associated with occupancy determination. 

This analysis suggests that in both 2016 and 2021, fewer people were imputed into dwellings correctly identified as occupied and non-responding than would have been expected based on the Post Enumeration Survey results. Overall, this under-imputation of persons has lessened nationally and for most states and territories between 2016 and 2021.

The Panel was impressed by the outcomes of the introduction of the ‘No-Census Number’ option. The option was developed to improve user-experience, increase Census response and coverage, and the Panel has concluded that this has resulted in increased data quality. Improvements to the coverage of and response from the population are evident through increases in the number of dwellings added to the Census Frame, as well as the use of this option by people staying in non-private dwellings, travelling, or in hard to reach locations.

Two new topics were added to the 2021 Census: Long-term health conditions and Australian Defence Force service. Their results generally aligned with the expectations from independent data sources, where available. The Panel observed that Census rates for mental health conditions were lower than expected compared with the 2017-18 National Health Survey, and rates of cancer were higher. The Panel concluded that more comprehensive and definitive data sources may be the best source of information for these items, with the Census advantage lying in data for small geographic areas and detailed cross-classification with other demographic characteristics. The item on Australian Defence Force service is also meeting a gap in veteran statistics. The Panel notes that ABS quality assessments and further evaluation are particularly important for interpreting and understanding these new data items.

In reviewing the 2021 Census data, the Panel identified some opportunities for future censuses that the ABS should consider:

1. There have been improvements to the Address Register’s maintenance and quality assurance procedures between the 2016 and 2021 Censuses. Continual improvements should be an ongoing focus, with a particular focus on missing dwellings. Results from the Census and Post Enumeration Survey should be reviewed for insights into any systemic gaps in the Address Register.
2. Response rates have improved overall for the 2021 Census, but there was a decline in some regional and remote areas of Australia. The Panel acknowledges that there may be logistical issues in conducting a census in regional and remote areas, particularly under pandemic conditions, and encourages the ABS to identify opportunities for improving response in these areas, many of which have had enumeration difficulties in previous censuses. Innovations in enumeration methods should be considered.
3. The ABS has made substantial investment over recent censuses in improving the count of Aboriginal and Torres Strait Islander peoples. However, the Post Enumeration Survey continues to show a high undercount for this population group. The Panel strongly encourages the ABS to review and investigate this issue with an emphasis on innovative solutions. In particular, the Panel considers that serious consideration should be given to the use of reliable administrative data to address missing responses to the question on Aboriginal and Torres Strait Islander origins. Any such change would need to be undertaken in partnership with Aboriginal and Torres Strait Islander peoples.
4. The Panel was pleased to note data quality improvements resulting from enhancements to occupancy determination and imputation for non-responding households, most notably as a result of the use of administrative data. ABS should continue to refine and enhance these processes. There have been notable improvements in reducing missing people within dwellings, and options could be investigated to further use administrative data to impute for other missing persons.
5. Similarly, occupancy determination was improved by the use of administrative data for non-responding households. The 2021 Census used electricity data in limited ways following the 2021 Census Administrative Data Privacy Impact Assessment. Further in-depth analysis of the level of improvement to census data from the use of electricity data at the household level should be undertaken, with a view to increasing its use in the 2026 Census. The Panel notes that this would be subject to a new privacy assessment based on the 2021 Census experience.
6. Innovative options introduced to the 2021 Census to enhance user experience and encourage response, such as the No-Census Number option, should be further developed for the 2026 Census.
7. Prior to the commencement of the consultation process on content for the 2026 Census, the ABS should carefully evaluate the two new topics, being Australian Defence Force service and Long-term health conditions, to inform the value of their continued inclusion in the 2026 Census.
8. The Statistical Independent Assurance Panel should be continued for future censuses to provide greater transparency, accountability and assurance.