PLANNING THE 1996 CENSUS
Well before one census is completed, work on the next census begins. Preliminary work began in 1990 on the 1996 Census, with investigation of alternative processing systems. In 1992 the development of the 1996 Census of Population and Housing began in earnest. The Government established an Interdepartmental Committee (IDC) in 1993, to consider options for reducing the cost of the census. The IDC recommended continuing a full census in 1996 and every five years
All aspects of the 1991 Census were examined by the ABS with a view to identifying areas of possible improvement. This included evaluation of form and question design, effectiveness of collection and processing systems and the quality and usefulness of the data provided to users. The major improvements to census procedures since 1991 are changes affecting mapping, collection management and output delivery.
The quality of the maps used by census collectors is an important factor in the census and, for the 1996 Census, the ABS is implementing a computer-based mapping system which will result in maps of much higher quality than previously. The digital map data used in the new system is supplied by the Public Sector Mapping Agencies. As well as improving the maps used by the collectors, the same map base will be used for the production of output maps.
The Collection Operation Management System (COMS) is a computer application developed in Lotus Notes by ABS staff for the 1996 Census. COMS will be used by 146 Census Field Managers to undertake the management duties of their position. This will result in a significant reduction in the clerical workload of Field Managers compared to previous censuses, allowing them to concentrate more on training and quality issues associated with the census collection.
For the first time all census processing and tabulation work will take place on microcomputers linked through a local area network. Microcomputer technology allows for design of systems which are generally more user-friendly and faster than is possible in mainframe environments. As a result, the processing system has been redesigned in a way which will contribute to improved accuracy of coding and enable the release of a wide range of final census data earlier than has been possible from previous censuses. As well, the use of a microcomputer-based tabulation system specifically designed for the census provides a timely and flexible service for users requiring customised tabulations from the census.
HOW IS CONTENT DETERMINED?
Consultation with the people who use census statistics commenced on the basis that the cost of the 1996 Census be no more than the cost of the 1991 Census.
In line with this policy, the ABS published in February 1993 a document outlining its initial views on the content of the 1996 Census, based on knowledge obtained from the topic submissions of the previous censuses and on known uses of census data. Users of census data and the general public were invited to comment on the views expressed in the publication.
More than 280 submissions received
More than 280 written submissions were received and these were followed up by consultation with major users and other interested parties. Some nine consultation meetings were organised in capital cities to discuss user comment and requirements.
In light of these consultations, the ABS proceeded to draft final recommendations on the nature and content of the 1996 Census. These were discussed with the Australian Statistics Advisory Council prior to the preparation of a submission to the Government. Census content was also discussed by the IDC, convened to review the cost of the census. On 6 September 1994 the Government announced its decisions on the 1996 Census, and documents outlining the content and the procedures for the conduct of the census, along with relevant regulations, were tabled in Parliament.
The Census of Population and Housing collects information by self-enumeration. Each household is asked to fill in the details specified on the census form with relatively little assistance from the census collector. Self-enumeration and the need to ensure that the large census operation is conducted as efficiently and effectively as possible, impose constraints on the types of topics and questions that can be included in the census if the public is to provide accurate data. There is also the need to limit the total number of questions asked in order to minimise the reporting load on households and census costs.
These constraints were reflected by the criteria used to assess topics for inclusion in the 1996 Census. These were:
whether the topic was of major national importance;
whether the census was an appropriate method of collecting data, given alternative sources and methods; and
whether the topic was suitable for inclusion in the census.
WHAT IS ON THE 1996 CENSUS FORM?
The consultation process led to the selection of three more topics for the 1996 Census: address of usual residence one year ago; issue (number of children ever born) and other dwelling (classification). All topics asked in the 1991 Census were retained, and further topics were investigated but did not meet the criteria for inclusion.
To decide on the layout and question wording for the census form, a program of tests is conducted before each census. For the 1996Census, a series of eight tests, including a dress rehearsal, were carried out for this purpose in various cities and rural locations between 1992 and 1995.
On Tuesday, 8 August 1995 a census dress rehearsal was conducted in parts of Sydney and Bathurst in New South Wales, and the Katherine region in the Northern Territory, to test collection and processing procedures for the 1996 Census.
Recording the answers
As in the 1991 Census, optical mark recognition (OMR) technology and computer-assisted coding will be used to capture the information from the forms. Most questions will again be answered by householders making horizontal marks on the form, which indicates their choice of answers from sets of alternatives. OMR technology allows these marks made on the form to be read by machines. Computer-assisted coding is an efficient method of coding the handwritten answers on the forms.
WHAT WILL IT COST?
The 1996 Census is expected to cost around $6.85 per person (at 1993 prices). Allowing for the effects of inflation, this is 2.2% less than the per capita cost of the previous census.