1001.0 - Annual Report - ABS Annual Report, 2004-05  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 11/10/2005   
   Page tools: Print Print Page  
Contents >> Section 2 - Special Articles >> Chapter 4 - The History of Australian Censuses of Population and Housing

Chapter 4 - The History of Australian Censuses of Population and Housing

INTRODUCTION

For 100 years the Australian Bureau of Statistics, and its predecessor the Commonwealth Bureau of Census and Statistics, has provided a national statistical service to the Australian, State and Territory Governments, Australian businesses, educational institutions, and the Australian community as a whole.

The type of statistical service provided, and the way in which it has been delivered, has changed markedly over the past 100 years. This change can be well illustrated by considering the evolution of the population censuses in Australia.

THE EARLY YEARS — COLONIAL STATISTICS

Australia has a history of regular population stocktakes from the time of the first British settlement. Arthur Phillip, in his reports back to England, included statistical information such as population numbers, tables relating to livestock in the settlement through to reports on the soldiers and provisions, and a general report on the sick and the number of deaths since landing in Australia.

In the early days of the colony Governors were required to report annually on the number of births and deaths, and records from population ‘musters’ exist for almost every year from 1790 to 1825. The first formal census of the modern type was held in New South Wales in 1828. It had been recognised that the previous proclamations by the Governor calling free citizens to muster had no legal force, and this census was authorised by an Act of the New South Wales Legislative Council, 9 Geo. IV., No. 4, dated 30 June 1828. What was distinctively new in this census was the first distribution of printed forms by responsible persons ‘by whom, as well as by respective Householders, who can write, each Form is to be signed when duly filled up (endnote 1).

The mainstream of official statistics in Australia began with the Blue Book. This was the annual report required by all Australian colonies back to the Colonial Office. When self-government was obtained in 1855, the early colonies — New South Wales, Victoria, South Australia and Tasmania — no longer had the obligation to produce statistics to meet the requirements of the Colonial Office and as a result statistical information began to diverge in both content and timing.

For the 1891 Census the colonies agreed on a common day, on a common core to the schedules, and on the compilation of the returns on a common principle. This was an important achievement, and it meant that a major stumbling block for uniformity in the 1881 Census, forming a common occupational classification, had been overcome.

STATISTICS FOR THE NEW NATION

The Australian statistical landscape, prior to and immediately following Federation, was coordinated by frequent Conferences of Statisticians, involving the Statists (the Heads of the State Statistical Offices). The meetings were to discuss statistical issues and agree on measures to aid in the consistency of statistics across the states and it was through these meetings that some of the Statists were beginning to see the advantages in the prospect of a national statistical authority that might lend its weight to the decisions of the conferences.

Federation was seen as a catalyst of statistical uniformity. In framing the Australian Constitution, the founding fathers had given the Commonwealth Government of Australia ‘... power to make laws for the peace, order and good government of the Commonwealth with respect to: ... (xi) census and statistics’. The Statists all agreed the new nation should have a responsibility and a capacity to undertake statistical inquiry.

The first major statistical business of the 20th Century was the 1901 Census. In March 2001 a Conference of Statisticians, including a representative from New Zealand, was held in Sydney to arrange the uniform collection of the 1901 Census. A concerted effort was made between the states to make the 1901 Census fully compatible between colonies. Unfortunately effort to achieve uniformity was more concerted in some colonies than others, and therefore was not entirely successful. From the outset it was generally accepted that the population estimates would be used as a basis for apportioning funds to, or for, the states. However the method of determining the final estimate eventually came into question, with each state proposing a different approach. This became the major problem associated with the 1901 Census. Following another two Conferences, the states eventually agreed on a uniform basis for determining the population in 1903.

The Australian Federal Parliament passed the Census and Statistics Act 1905 (C&S Act) on 8 December 1905. The C&S Act stated that a census should be taken ‘ … in the year one thousand nine hundred and eleven and in every tenth year thereafter’. At a conference of State and Commonwealth Ministers in the following year, it was resolved ‘the general statistical departments should be handed over to the Commonwealth’ and the Commonwealth Bureau of Census and Statistics was formed.

Image: 1911 Census Proclamation
1911 Census Proclamation


The first Commonwealth Census was undertaken in Australia at midnight between 2 and 3 April 1911.

Two small cards were required to be completed for the 1911 Census — a household card and a personal card. The census personal card comprised 14 questions for the householder, and the household card consisted of a further five questions. Particulars collected included name, sex, date of birth, marital status, current marriage date, number of children, relationship to head of household, blindness and deaf-mutism, birthplace, nationality and race, length of residence in the Commonwealth, religion, education, occupation, type of building, building materials, number of rooms, and owner/renter status.

The 1911 Census attempted to measure attainment of university qualifications. Unfortunately the quality of data received on this topic was so poor that the information was never released. It was 1966 before another attempt was made to obtain this information.Tabulation of census data was undertaken in Melbourne and completed almost entirely by hand. Staff were required to physically sort more than four million cards and count them for each tabulation.

Results from the 1911 Census took a lengthy time to release, and were further delayed by the commencement of World War I. Three volumes of results were planned. Volumes II and III were released in September 1914, however Volume I (which included the Statistician’s Report) was not released until 1917.

The second national census was held at midnight between 3 and 4 April 1921. The content of the census was similar to the first, and the topic ‘cause of unemployment’ was added to the schedule. Based on a recommendation from the Victorian Statistician, staff from state Electoral Branches were used to collect and distribute the 1921 Census forms.

The data was tabulated using automatic machine tabulation equipment for the first time. Hollerith machinery was hired from England for this purpose. Seven months after the census date, basic data from the 1921 Census were released in bulletins, the first being a count of the population of the states and territories. The Statistician’s Report and a full summary and analysis of the census was not released until 1927.

By 1930 it was apparent that Australia was in the midst of a major Depression and a decision was made to delay the scheduled 1931 Census, and to do so meant a change to the C&S Act was required. In 1930, the C&S Act was amended by the addition of the words ‘ … or at such other time as is prescribed’. Ultimately, the need for information on the impact of the Depression, as well as the realisation that the census would stimulate economic activity by providing employment and business opportunities, resulted in government approval of a census to be held in 1933.

The 1933 Census was the first census after the Bureau moved to Canberra, and as such it was the first to be tabulated in Canberra. New topics on industry, orphanhood, foreign languages, war service and income were included.

Updating the detailed maps of all areas of Australia is a by-product of a census. For the 1933 Census, the compilation of maps took around nine months and employed 60 survey draftsmen.

The census form was a single schedule, with the particulars for each individual in a household listed. Dimensions of the new form were large and required triple folding to obtain the size of a foolscap sheet. Personal slips were available for those who requested them.



THE POST-WAR AGENDA

The fourth Commonwealth Census was expected to be in the early 1940s, however it was delayed because of World War II. At the end of the War the Bureau commenced working towards a new census, to be held in 1947.

Several of the questions from the 1933 Census were removed and a number of new questions included. The 1947 Census marked a major change in the Bureau’s assessment of employment data items. In the published results the Bureau departed from the long tradition of publishing data on ‘bread-winners’ to embrace the new concept of ‘economically active’ or ‘labour force’. Retired men on independent means were previously classified as ‘bread-winners’, but in the 1947 Census this group was classified as being ‘not economically active’.

In 1954, for the first time, the processing of the census was decentralised with three processing centres instead of one. The processing centres took varying times to complete their work, and there was some evidence they were not consistent in their coding. This finding made the Bureau wary of decentralised processing for future censuses. Also for the first time, international organisations were mentioned in the Statistician’s Report as having an influence on the development of the Australian Census. In particular the Statistical Office of the United Nations was commended in relation to the development of uniform standards, definitions and procedures.

The 1954 Census marked a major change in the use of punch card operators, with the introduction of Hollerith Mark Sense Gang Punches. They automated card punching, and staff that could mark the cards largely replaced skilled punch card operators. These were then punched automatically as the marks were read using an electrical conductivity sensing technique. Until this census there was a strong demarcation between male coding jobs and female punching tasks and in 1954 women began to be used to code the data in substantial numbers.

Sampling techniques were used for the first time for the 1954 Census to obtain data on married couples and their families.

BIG IDEAS AND NEW BEGINNINGS

The 1961 Census put timing back on track for a decennial (ten-yearly) census and paved the way for quinquennial (five-yearly) censuses. Additions to the schedule included: questions on qualifications for current occupation; state of usual residence for those temporarily absent from their homes; the existence of a television in the home; and for those living on farms, the distance to their local post office and the size of their holding. A question on qualifications was used to assist in coding occupation data. A question on state of usual residence appears to have been the Bureau’s first attempt at obtaining some form of de jure measurement of the population (according to the place of usual residence rather than the place of enumeration).

The Australian Classification of Occupations used in the 1961 Census was adapted from the principles embodied in the first International Standard Classification of Occupations, published in 1958 by the International Labour Organisation.

The Bureau was eager to centralise data processing once more for the 1961 Census because of coding inconsistencies in the 1954 Census. The size of the workforce, as well as the machinery, required a larger space than the Bureau was able to rent on a short-term basis, therefore two centres were utilised in Sydney. The coding inconsistency problem was avoided as all coding was processed at one site while the second site undertook preliminary data editing. The cards were then sent to Canberra for final tabulation.

The 1966 Census was one of many firsts:
  • the first census to be held five years after the previous census
  • the first time the number of questions on the census schedule increased substantially (from 15 in 1961 to 24 in 1966)
  • testing of the schedule was introduced for the first time
  • the first time a computer was used for processing the census
  • quality control methods were introduced to help ensure the accuracy of census coding
  • 'community profiles’ were introduced – a standard set of tabulations covering most census topics for every geographic area from individual collection districts to the whole of Australia. Community profiles remain the most popular tables produced from the census data to this day
  • microfiche replaced paper publications for the majority of census tabulations
  • tabulations were made available on magnetic computer tape
  • clients could request tailored tabulations of census data
  • thorough coding of family structures
  • post-enumeration surveys were introduced with the aim of checking the accuracy of census figures. These surveys provide an estimate of the undercount of a census which has become a key measure of the success of Australian censuses.
In 1967 the Commonwealth Government held a referendum that resulted in the removal of section 127 (as well as part of another section) from the Australian Constitution. This section had been the barrier to the Bureau including counts of Aboriginal people (but not Torres Strait Islanders) in published census figures.

Theoretically this should have simply resulted in the addition of the number of Aboriginal people to population estimates as the Bureau had been counting Aboriginal people, though not including their numbers in the published results. However consideration of the treatment of the counts on Aboriginal people proved problematic. There was also a difficulty with the question. Previously the Bureau had attempted to identify ‘Aboriginality’ from the race question, which included asking people to identify what proportion they were of various ‘races’. In 1971 the Bureau completely redesigned the question on ‘race’, and identification for people of Aboriginal or Torres Strait Islander origin was requested.

In 1973 the Whitlam Labor Government established the Committee on Integration of Data Systems known as the Crisp Committee after its Chairman, L.F. Crisp. The Government had been concerned about recent discrepancies in statistics from various Commonwealth departments and the lack of statistical data on key areas of the economy, and believed this could interfere with its reform agenda. As a result, the Committee ‘undertook wide ranging investigations of Australia’s statistical system’ (endnote 2).

The Crisp Committee reported in March 1974. It recommended the establishment of the Australian Bureau of Statistics as the central statistical authority with full statutory powers, administratively independent of any department and thereby perceived to be policy neutral. The Australian Statistician was to be a statutory appointee vested with the powers of a head of department under the Public Service Act 1975. This led to the development of the Australian Bureau of Statistics Act 1975. Under this Act the Australian Bureau of Statistics was established with the role of the central statistical authority for the Australian Government and, by arrangements, for the governments of the states and territories.

The 1976 Census was developed during a period of great change in Australia’s history. The Whitlam Government was eager to see statistics to support the major social changes it envisioned for the country. The number of personal questions increased to 41 and the dwelling questions to 12. The new topics went before both Houses of Parliament and were passed without objection from the Opposition, which held control of the Senate.

In the two months prior to census day there was considerable public debate about privacy and the census. The concerns were regarding the nature of some of the questions asked and whether the respondent’s name was necessary on the schedule. The Treasurer requested the Law Reform Commission to investigate privacy and the census and to make any recommendations it thought necessary to improve census privacy. This resulted in a major investigation for which the Bureau had to justify its census processes and questions.

The census was conducted on 30 June 1976. Preliminary checking and processing were conducted in the states, with the main processing again taking place in Sydney. Only basic data on age, sex, marital status and birthplace could be tabulated as budgetary constraints meant the Bureau could not proceed with processing the rest of the information collected, and only 50 per cent of the 1976 Census forms were processed. The post-enumeration survey established there had been a higher under-enumeration for the 1976 Census than for previous censuses.

Up until the 1976 Census the population estimates were based directly on the census population counts. From 1976, these counts were adjusted for under-enumeration as measured in the post enumeration survey to determine the estimated resident population. Australia is believed to be the first country to do so formally, although others have since followed.

In light of the privacy concerns expressed in the lead up to the 1976 Census, the Law Reform Commission released its report in 1979 on Privacy and the Census. The report endorsed the importance of the census and agreed that the processes used were appropriate, including the requirement for householders to supply their names on the form. Many of its recommendations were adopted by the ABS, including one that there should be an intensive advertising and publicity campaign to explain the census and the measures taken by the ABS to protect confidentiality. Perhaps of even more importance was a change by the ABS to a more open and public approach in developing each census.

THE BEGINNINGS OF PUBLIC ENGAGEMENT

The 1981 Census saw more public consultation than any previous census. From late in 1977 the Bureau advertised in major newspapers seeking public submissions on the content of the census, as well as approaching Australian government departments, state departments and civil liberty organisations. Over 1,600 submissions were received for topic inclusions (with 40 new topics suggested) and only 60 or so for topic exclusions. Questions on income, one of the more controversial topics in the 1976 Census, received a great deal of support.

The Bureau began to focus on encouraging a greater understanding of the value of census data in the population at large and in the 1981 Census there were 31 questions.

This census saw the development of new procedures to enumerate Indigenous Australians. A special census form was developed for Indigenous peoples located in remote areas. For the first time new services were made available for other ethnic groups including multilingual brochures and telephone interpreters.
Data from the 1981 Census were made available on maps, microfiche and magnetic tape as well as the usual publications. Social Atlases were produced for the first time.

THE MODERN CENSUS ERA

Significant change in the collection procedures was undertaken in the 1986 Census. Since the 1921 Census, the Australian Electoral Commission had been used to organise and supervise the distribution and collection of census forms. For the 1986 Census the Bureau took over the management of the collection in New South Wales and Victoria due to industrial relations issues occurring in those states at the time. This proved successful and the Bureau decided to adopt this procedure for future censuses in all states. A whole new field collection system was developed.

In 1983 amendments were made to the Census and Statistics Act 1905 requiring that all topics included in a census were to be prescribed by a regulation. Previously there was a list of topics in the Act that were required to be asked, and any other topics had to be prescribed by regulation.

The 1986 Census saw a range of new media for data releases. In particular diskettes and CD-ROMs were adopted as vehicles for data release. The main feature of the CD-ROM product was the combination of census data with mapping data and software so that clients could easily map the data.

The census day was moved from 30 June to 6 August for the 1991 Census to avoid all school holiday periods. For the first time each ABS regional office took over the management of field operations from the Australian Electoral Commission. This census also marked the start of regular consultations with the Office of the Federal Privacy Commissioner on operational procedures.

Processing of the 1991 Census used optical mark recognition to capture information on the form, significantly reducing the time required for data entry. Computer assisted coding was used for the questions which required coding. This had the dual benefits of reducing the manual coding load and improving coding consistency.

This census also saw the first edition of Census Update, released in September 1991, with the aim of maintaining a regular flow of information to clients on the census. The publication is still released periodically today.

In 1993 an Interdepartmental Committee was asked to consider options to reduce the cost of the census. The two possible alternatives — reducing frequency or reducing content — were found to be ‘not feasible’ and the Committee recommended that a five-yearly, full content, census would continue.

In the lead up to the 1996 Census there were murmurs of concern about privacy issues through the community, reminiscent of the 1976 Census. However by census day media opinion appeared to have swung back in favour of the census and few complaints were raised in the media over privacy.

A major change for the 1996 Census was the use of Geographic Information Systems to generate the printed census maps for census collectors which involved using a single electronic map database that covered the whole of Australia.

The most significant change for the 1996 Census was in the release of the data. In a two-phase release process, with those topics that could be easily processed released first. The majority of results were released by 15 July 1997, less than twelve months after the census, and the Adelaide Advertiser described the 1996 Census as ‘a beautiful set of numbers’ (endnote 3).

In 1998 the Australian Government decided that people would be given the opportunity to have their personal details, with name identification, preserved for release in 99 years by the National Archives of Australia; an initiative of the Centenary of Federation celebrations. Slightly more than 50 per cent of the Australian population chose to opt into the ‘Time Capsule Project’.

Reflecting the expansion of new technologies, questions on the use of personal computers at home and of the Internet were added to the 2001 Census.

The 2001 Census was the first to use intelligent character recognition that could capture both tick box and textual information from the census form, allowing most of the textual information to be automatically coded. Images of the census form were used during processing rather than the paper census forms. For the first time also, the World Wide Web became a key part of the data release strategy including release of the community profiles.

Detailed results from the 2001 Census were the quickest release ever of an Australian census; 11 months after census day.

CONCLUSION — THE FUTURE OF THE AUSTRALIAN CENSUS OF POPULATION AND HOUSING

Increased awareness of the importance of the census for the community has steadily grown over the past 100 years. As the desire to understand more about the community in which we live grows within our society, and governments at all levels embrace ‘evidence based planning and policy’, so does the significance of the census.

Advancements in technology will result in some significant changes in the future. Developments in technology for Internet based data collection, for those who choose to respond in this way, are underway. This has the potential to significantly reduce census costs as well as improve the quality of the data collected. While it will be possible to complete the 2006 Census form on the Internet, it is in the long term that major cost savings will be realised. Advancements in technology will also change the means by which census data can be disseminated to the users of this information.

Changes in topics will continue to occur to reflect the statistical needs of the Australian, State and Territory Governments and the community as a whole. Four new topics will be included in the 2006 Census, incorporating questions on: the number of children ever born; the need for assistance (disability); unpaid work; and household access to the Internet.

The ABS will continue to look for innovative ways to improve the statistical usefulness of census data for the Australian community as a whole; for example, the Census Data Retention project and the Census Data Enhancement project. The Census Data Retention project provides the opportunity for Australians to have their personal details preserved for release in 99 years from census day. As previously offered in the 2001 Census, this opportunity will be made available in the 2006 Census and in all censuses thereafter. In Discussion Paper: Enhancing the Population Census: Developing a Longitudinal View (cat. no. 2060.0) the ABS proposes to enhance the value of the 2006 Census data by combining it with future censuses. In recent years there has been a recognition of the importance of data to help understand how people move through changes over time and what factors influence these changes. In its existing form the census provides a ‘snapshot’ about Australian people and households once every five years. The creation of a Statistical Longitudinal Census Dataset (SLCD), combining census data over time, would provide the means to identify patterns of change in social and economic circumstances for individuals and households over time, and the increased information would enhance decision-making processes. The SLCD will be based on a 5 per cent sample using statistical matching techniques. Name and address information will continue to be destroyed after publishing as has been the case for previous censuses.

The Census of Population and Housing continues to be the most detailed and important source of statistical information for small geographic areas and population groups produced in Australia. In a world where information abounds and there are often several sources for similar information, there is still no source that comes close to competing with the fine level of detail available from the census.

ENDNOTES

1 G.H. Knibbs, The Private Wealth of Australians and its Growth as ascertained by various methods together with a report A Report of the War Census 1915, CBCS, Melbourne, 1918, pp. 8–13, 19.

2 Miller, John, 2000 Interview.

3 Adelaide Advertiser, 17 July 1997



Previous PageNext Page