1504.0 - Methodological News, Sep 2011  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 28/09/2011   
   Page tools: Print Print Page Print all pages in this productPrint All

Industrialisation of Statistical Processes, Methods, and Technologies

At last month's International Statistical Institute 2011 World conference, the ABS presented a paper on the topic of the industrialisation of statistical processes, methods and technologies. This paper was a collaboration between IMTP (Geoff Lee) and MDMD (Lisa Apted, Philip Carruthers, Daniel Oehm and Frank Yu) and discussed the industrialisation of statistical activity in national statistical agencies such as the ABS. The paper illustrated three examples within the ABS where industrialisation might be usefully applied. This article briefly summarises the paper.

During the industrial revolution, so-called "cottage industry" approaches were replaced by standardised processes, new methods, and innovative technologies which had far reaching impacts on the efficiency and effectiveness of manufacturing production. The revolution created an environment in which outputs could be delivered more easily and cheaply. The outcome was not only a reduction in the resource base required but also an outpouring of new possibilities for utilising the products made possible by their increased availability at reduced cost. The "industrialisation" concept of standardised production processes -- where new production methods and technological advances allowed production on a large scale -- worked in a manufacturing context, and there are sufficient similarities to suggest it should work equally well in a statistical production context. Here, it is assumed that cheaper access to greater quantities of high quality statistics will encourage the generation of more innovative presentation, reporting and analysis methodologies.

In the official statistical context, it is important to understand that industrialisation does not represent the robotic automation of standardised statistical activity. Rather, it offers improved capacity to replicate basic processes, freeing up analyst resources so they can add value where needed, with the support of knowledge-based decision-making tools.

There are many different statistical processes within the ABS that are amenable to greater industrialisation, including the seasonal analysis of time series, microdata confidentialisation and the compilation of price indexes. Standardised metadata can be used to drive business processes and capture data relationships to promote greater harmony between similar or dependent activities. Since at their core all of the production processes for official statistics are about manipulating and quality assuring information, information management must be a strategic focus to achieve greater harmonisation and "industrialisation".

Anyone who has been involved in the process of developing an international statistical framework will be aware that it can be a slow and tedious process. Reducing the rate of innovation in the production methods of official statistics is a genuine risk to the "industrialisation" philosophy. It need not necessarily be so, as the experience of the growth of the internet has shown. The internet has created an environment within which enormous bursts of technological creativity and innovation occur - and yet at its heart it is reliant upon some basic standards which have been widely adopted. The task for methodologists working in the production of (official) statistics therefore becomes one of establishing which methods can and should be standardised, and which should not.

For more information, please contact Philip Carruthers on (02) 6252 5307 or philip.carruthers@abs.gov.au