ABS is planning a major change process aimed at improving our management of statistical information, from the initial data collection, through the processing cycle to its availability as a coherent, accessible, and well managed set of statistical outputs. We have been exploring what is required to achieve transformation in our operations and in our transactions with our users. We have realised that focussing on information management is necessary but not sufficient. We need a coordinated program of changes to business process, technology and methodology as well as to information management policies and practices.
In discussion with other NSIs, we have recognised that the general problem is to harmonise processes, methods, capabilities and technologies. In other words, to “industrialise” the whole statistical production process, away from cottage industry towards large scale, repeatable approaches.
This implies we should seek:
general formulae that would provide a rich set of solutions suitable for different situations;
methods and tools which are designed for automation, for repeatability, and for large scale parallel operation;
processes which handle complex data structure and output needs, involving multiple data sources and complex relationships connecting datasets;
ways of managing knowledge and information which support repeatability, and multiple parallel use, continuous improvement and optimisation.
The question then becomes, how does this apply to the methodology development and implementation practices in NSIs?
Using some practical examples in seasonal adjustment, editing, and microdata confidentialisation, we explore the challenges and implications that “industrialise” may have for the methodologists in an NSI.
Of course, industrialisation is about achieving economies of scale, and efficient use of resources, and the obvious extension is to ask whether industrialisation of methodology would offer advantages if extended internationally, across many NSIs. Using some examples from an international collaboration project we explore how the future methodological infrastructure may be efficiently developed.
Keywords: Statistical information; Data collection; Industrialise
Biography: Since February 2010 Geoff has been the Head of the Information Management Transformation Program, which has responsiblity for modernisation of the information management infustructure within the ABS to support its data management and business process standardisation.
Prior to this Geoff was the First Assistant Statistician heading the Methodology and Data Management Division which has responsibilities for supporting the statistical collection process, by providing advice on survey design, methods and on data quality through all stages of the survey cycle.
Geoff has worked in the ABS since 1979, spending most of his time in this Division. Geoff's interests and expertise cover all aspects of methodology - sample design and estimation, survey processing, detection and treatment of non sampling error, seasonal analysis, data confidentiality protection, quality assurance and improvement, and training of mathematical statisticians and analysts.