• Summary
• Downloads
• Explanatory Notes
• Related Information
• Past Releases

The 2001 Census data quality information was released in two main formats, Census Fact Sheets and Census Papers. Census Fact Sheets were brief reports containing non-response rates and other information on data issues as they arose and Census Papers were more in-depth investigations of quality issues and the processing methods used for a range of topics. These were released progressively some time after the release of Census data. The use of the 2001 data quality products has increased significantly with their placement on the ABS web site, however consultation with users about their needs for 2006 Census information indicates improvements can still be made.

The information paper User Views on 2006 Census Output Strategies (cat. no. 2009.0) and the accompanying questionnaire included sections on data quality measures and information on how the ABS and the Census program are currently reviewing the sort of data quality information provided and, just as importantly, how users can more readily be made aware of it and access it. Feedback from the user survey highlighted the importance of information in the areas of:

• accuracy
• consistency
• comparability
• accessibility
• non-response
• under and overcount measures.

• effectiveness of ABS enumeration strategies, especially in remote areas
• effects of processing and editing on the data
• imputation methodology
• confidentialisation process
• comparability of Census data to other ABS data sets.

For 2006 dissemination, it is proposed to make basic data quality information such as non-response rates available with data release, and to provide different levels of detail through linked files on an on-going basis. In this way, users can access the level of data quality information they want, and information about data can be added as required or when issues are identified.

The following outlines the proposal in some detail:

1. At Census data release, each variable selected in a table specification will be able to be linked to the corresponding entry in the Census Dictionary, to the relevant classification, to the non-response rate if applicable, and to any quality issues already identified.
2. The next level would be a variable specific quality statement covering - if and where relevant - such things as coding discrepancy rates, processes and edits applied to the data, imputation rates, undercount information, as well as information about comparing Census data to other data sets and outlining the reasons for apparent differences. Some of these reports may be available at the same time as data release and others, depending on the nature of investigatory work required, a short time later.
3. Aspects of Census operations that are not necessarily linked to any particular variable, or particular topics or issues identified by user interest, would also be made available through the dissemination Web pages. These may include enumeration strategies, the quality management strategy, editing strategy, imputation methodology, and reports on automatic processing performance and other technology used in the program.

In summary, the proposed 2006 Census data quality output should provide more information earlier, at different levels of detail, and be readily accessible through links on the Census Web pages. In-built functions of these pages will provide the ability to print or download the information along with the data.

You are welcome to comment on these proposals through the questionaire provided with this publication or you can discuss this further with Joanne Healey on (02) 6252 5611 or email at joanne.healey@abs.gov.au.