1504.0 - Methodological News, Dec 2014  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 17/12/2014   
   Page tools: Print Print Page Print all pages in this productPrint All

Methodology Architecture – A Roadmap for New Methodological Directions in the ABS

To maintain its strong brand as a central statistical agency into the 21st century, the ABS is transforming the way it acquires, collates, uses, reuses and disseminates statistical information. To support this transformation, innovative, industrialised and contemporised statistical methods and tools will be required. Methodology Architecture provides a roadmap for systematically assessing and developing these 21st century statistical methods and tools covering the full spectrum of the statistical production cycle.

The ABS transformation vision comprises a products vision and a process vision. The ABS’ products vision is a 21st century National Statistical Organisation (NSO) which has the ability and agility to combine ABS data with other strategic sources of data, including administrative data, transactional data, Big Data, and “organic” data, to produce more timely and relevant official statistics. As well, the process vision sees the ABS fully embracing industrialisation and standardisation in its business model in the production of statistics.

The Methodology Architecture vision for the ABS is “to provide a set of methods that underpins the products and process visions of the ABS transformation program.” Fundamental to achieving this vision are five key elements in developing the future inventory of statistical methods and tools: innovate, industrialise, contemporise, build capability, and build support.

Innovate: New methods and tools are required to address the new and emerging challenges from the ABS products vision. An example of this is in the linking of data to create fused unit record files or new statistics, and their processing, analysis and dissemination at the micro and macro data level. The goals are to improve relevance, production cycle time, organisational capability and meet legislative requirements to protect the confidentiality of personal or business information.

Industrialise: In the post transformed world, we see significant improvement in the production cycle time through re-use of data available within and without the ABS, as well as through re-use of statistical processes, statistical methods and tools. For this to occur, the post transformation methods and tools need to support plug and play, standardisation and corporatisation, being connectible and metadata driven, and being user-driven.

Contemporise: The survey methods used by the ABS are predominately design based (often model-assisted) and in time series analysis predominately filter based (X11, X12ARIMA) methods are used. There is an aspiration for more use of statistical models to guide our choices of estimators in our methods. In addition, responsive design methods are starting to be used to improve the cost-efficiency of survey collections and reduce survey errors.

Build Capability: The ABS framework for building methodology capability is based on attracting the best budding methodologists from university graduates, offering comprehensive training to graduates, offering interesting, challenging and varying work as well as providing opportunities to attend conferences in order to retain staff, peer reviewing key methodological work and collaborating with several universities for research, technical advice and teaching short courses.

Build Support: It is as important to build support from the senior managers of the ABS as from methodologists for the methodology architecture. We found it very useful to establish a small team of champions who are enthused about the new methods, expose them to the new techniques, and encourage them to promote these new ideas to their colleagues.

To be successful, the Methodology Architecture must address all 5 key elements satisfactorily. Each of the elements poses its own challenges, and we will work with academics, experts in the field, other NSOs as well as ABS stakeholders, including methodologists, to address them.

A more detailed version of this article is to appear in the next edition of the Statistical Journal of the IAOS.


Further Information
For more information, please contact Siu-Ming Tam (02 6252 7160, siu-ming.tam@abs.gov.au).

The ABS Privacy Policy outlines how the ABS will handle any personal information that you provide to us.