The Integration of Statistical Processes through IT Support, Methodological Support and Standards Development
Marlize Pistorius, Jairo Arrow
Methodology & Evaluation, Standards, Statistics South Africa, Pretoria, Gauteng, South Africa; Methodology & Standards, Statistics South Africa, Pretoria, Gauteng, South Africa

This paper presents the architectural concepts, tools and the methodology used for the implementation of an integrated information system for Statistics South Africa (Stats SA). The paper focuses on the integration between systems and standards developments and methodological support during the design and implementation of statistical production processes.

The focus will be on standardised sampling methodologies for both household and business surveys. A system design which can be either generic and a metadata driven database or a hybrid method between a normalised and metadata driven database will be examined.

An array of integrated IT systems tools are used to support and optimise the underlying statistical production processes. During the analysis phase data stored in SQL2005 are made available to the Statistical Analysis Software (SAS)® platform. The SAS® repository provides the ability to access data from all data sources from one application while being able to perform data analysis and publish from the same application. The entire SAS® platform runs on Virtual servers which ensure that access to the server are always guaranteed.

Edit and imputation rules are captured with a decision-table tool LogicGem. The rules are then converted to SAS code. Through the use of SAS Enterprise guide generic methodologies can be applied. Imputation, editing and non-response adjustments are carried out with help of CLAN. Statistical Macro Extensions (StatMX) is used for weighting and estimation. Finally certain Microsoft Office Add-Ins are used as an part of creating statistical publications. Dissemination of management information is allowed through the Web-Tier which is a presentation layer and it allows users to explore data.

The potential of integration can only be exploited through a holistic approach focusing on the integration of statistical data and the synergy of statistical processes using information and technology.

Keywords: SAS®; LogicGem, CLAN, StatMX; Integration; Web-Tier and sampling methodology

Biography: Marlize Pistorius is part of the senior management of Statistics South Africa. Marlize heads up the Methodology & Evaluation as well as the Standards divisions within Statistics South Africa. Part of her responsibility includes system development, providing methodological support for Economic Statistics and Household based statistics; as well as Evaluation within the organisation which includes the execution of the Post Enumeration Survey of the Census 2011. One of her areas of responsibility also includes development of Standards to ensure that coherent and quality statistics are produced.