QUALITY MANAGEMENT IN THE ABS
The ABS's strong commitment to maintaining and improving the quality of its statistical program is reflected in its Mission Statement 'to assist and encourage informed decision making, research and discussion within governments and the community by providing a high quality, objective and responsive national statistical service'.
The development of appropriate service charters provide a clear articulation of ABS quality service objectives. The ABS currently has two such service charters. The Client Service Charter describes the relationship between the ABS and users of its products and services. The Business Surveys Charter sets out the relationship between the ABS and businesses that provide it with information for statistical purposes. Both Charters include performance standards for the relationships between the ABS and its clients, and its service delivery. Copies of the Charters are available on the ABS website and through bookshops located in each ABS office.
Making Quality Visible (MQV) has been a key focus of the ABS since the late 1990s. The main tenet of MQV is that the users of ABS statistics need to understand the quality of those statistics in order to judge whether they are suitable for purpose. A Data Quality Framework has been developed to provide a basis for Quality Declarations surrounding ABS statistics. Work done by Statistics Canada and the International Monetary Fund (IMF) were important inputs to the Australian development. More recently, the increased importance of web-based dissemination of statistics has introduced new challenges to MQV, and the ABS is currently considering options for Quality Declarations and complementary Quality Statements that can be linked to statistical outputs released in a web-based environment.
No single group within the ABS has responsibility for assuring the quality of ABS statistics. Rather, each ABS statistical area, including the area responsible for the compilation of the national accounts, is responsible for continuous quality review and improvement. Each subject matter and support area in the ABS provides a regular report to the senior management forum, highlighting progress and problems in their area and providing performance measures.
While the ABS does not have a centralised quality management system, the quality of ABS statistics is underwritten by the application of good statistical methods during all stages of a collection including the design stages. The ABS has a relatively large Methodology Division which is responsible for ensuring sound and defensible methods are applied to all collections and compilations. Also, the Methodological Advisory Committee, a group of academic experts, provides independent reviews of ABS statistical methods.
The ABS puts substantial effort into developing statistical standards, including concepts, data item definitions, classifications and question modules. Standard rules are adopted for survey frame maintenance, field collection and estimation, and generalised processing facilities are available to support the use of these rules. All ABS surveys must use these standards.
In the area of economic statistics, the international system of national accounts (SNA) is used as a coordinating and integrating framework for the collection of data. A dedicated area of the ABS has responsibility for the management of the integration process, and to ensure that appropriate standard definitions, classifications and collection units are used in all economic collections consistent with the system of national accounts. The national accounts system also provides an important basis from which relative priorities for the collection of data for the various aspects of economic activity can be assessed. The 'annual integrated collections' program of industry surveys are designed around national accounts priorities and user priorities for other data on specific issues.
At the output end of collections, each subject matter area is required to confront its data with other ABS data and with external information, to ensure statistical coherence. The key macro-economic statistics are "signed off" in meetings established for the purposes of clearing the statistics.
Key measures for demonstrating reliable/accurate statistics include high response rates, low sample errors and the timeliness of ABS statistics. A standard set of measures has been developed to present a comparison of quality across collections. The key indicators are also included in the ABS's annual report.
A further key element is the availability of information to enable users to make their own assessment of quality. Information on reliability and accuracy, as well as extensive information on the statistical methods used in collections are routinely provided in concepts, sources, and methods manuals, information and research papers and explanatory notes in publications and on the ABS website.
While assessing quality for any statistical collection is never easy, it is particularly difficult in the case of the national accounts. By design, the national accounts are compiled within a framework that integrates a wide range of statistics covering the whole economy. The accounts utilise a myriad of data sources, both internal and external to the ABS which become available at differing time periods. The integration of these diverse statistics is a complex task requiring the use of a wide range of methodologies, processes and procedures. The resultant datasets typically then undergo several revisions as more and better source data are incorporated into the system over time.
Because of these complexities, national accounts data cannot be subjected directly to error measures, such as sampling bias and variance, that are normally associated with statistical outputs based solely on survey methodology. This makes quality measurement for the national accounts an even more subjective process than quality measurement for outputs based on, say, a single survey collection. An overall view about the quality of the national accounts has to be built up from a staged assessment of the data and statistical process used to produce national accounts outputs. In assessing quality, the following three distinct phases in the processing cycle need to be considered:
- the input data
- the transformation of the input data into intermediate data
- the transformed output data.