|1. Introduction (Statistics South Africa)||Statistics South Africa||3. Statistical Metadata in each phase of the Statistical Business Process (Statistics South Africa)|
2.1 Statistical information model
2.2 Adoption of GSIM
2.3 Statistical business process model
Overview of the Process Model
Also shown in figure 6 are the main phases of Statistics South Africa's statistical cycle, internally referred to as the statistical value chain (SVC). The development of the SVC forms part of the organisation's standardisation of processes and systems. The SVC will provide broad guidelines for survey areas.
It was decided that the organisation should not re-invent the wheel in its development of the SVC. Therefore in developing the survey cycle, the starting point was to study how national statistics organizations (NSOs) similar to ours had logically structured their survey cycles. Another guiding principle was that Stats SA was not re-engineering the way it conducts surveys, but wanted to map and structure it as much as possible. We had to adapt survey cycles from organisations whose survey operations mimicked those of us. This work was done in 2005. Having made all considerations, we found the Statistics New Zealand's business process model for survey cycles to be the most aligned to our survey operations.
How Stats SA's Statistical Process Maps into Common Metadata Framework (CMF) Lifecycle Model
Resulting form the process stated above, Stats SA's survey cycle consists of the following phases:
Although Stats SA already produces statistics that satisfy needs of a large and diverse group of the country's socio-political landscape, requirements for new or supplementary statistical products often arise. Requests for such new products may emanate from a number of sources, including government departments, private business and other stakeholders. When this happens, the first step in statistical production is to understand the need for the required statistics, i.e., what the required statistics are going to be used for in concrete terms by their users. Often for new projects, a lot of detailed information about what is needed is not at first clear and in some cases, not present, even from the perspective of the initiators of the project. It is therefore important to go through a process of refining the understanding the information (statistics) needs to be addressed by the results of the project.
The design phase consists of preparing ground for the execution of a statistical production project. This stage is reached when either a new project has been given the go ahead or a frequent on-going project is about to begin. In the case of on-going surveys, the design is usually in place already, but in a few cases it might need to be altered to cater for additional requirements or special circumstances.
The build phase puts together all the pieces of the infrastructure for a statistical production project. These include the computer system, scanners, printing out of questionnaires, etc. The build phase also includes the procurement of the pieces of the infrastructure as necessary. The amount of work done in this phase also varies for new and on-going surveys.
Although the term collection may have different meanings in a statistical organization, it is used in the SVC to refer to both direct and administrative methods of data collection. The direct collection method refers to data collection in which Stats SA sources data directly from the respondents. In administrative collection, data are drawn from databases of other organizations which in turn source them from their respondents. It is important to note that this phase has a major influence on the activities of the design phase.
The Process phase includes capturing collected data into databases so that data processing may be done. Data processing is necessitated by a number of issues. Chief among these is a fact that the data collection process is fraught with errors. The process phase is undergone to remove these data validity errors so as to improve data quality, and to package the data for use by analysis tools.
After data have been cleaned during the Process phase, they are now ready for manipulation using analytical tools. This is the analysis done by domain experts to get insight into the meaning of the data. Further data quality enhancements may be done at this phase.
Stats SA collects data in order to produce statistics to be used by different stakeholders in the country including the general South African public. This means that the organization has to have ways of giving these communities access to the data and the resultant statistics. The Disseminate phase formalizes the steps Stats SA needs to go through in order to distribute information to the different communities as well as give them access to data repositories.
Stats SA uses a number of dissemination methods to ensure that the data produced by the organization is accessible to the widest user community. These include: electronic (e.g. via the internet), printed output and compact disks.
Table 1 below shows the mapping between Stats SA's survey cycle and the METIS cycle:
Survey planning and design
Need and Design Phases
Part of Design Phase
Derivation, Estimation, Aggregation
Post Survey Evaluation
Table 1: METIS Cycle vs. Stats SA's Cycle
Post Survey Evaluation is currently done outside the statistical cycle. It is performed only for the large surveys such as the population census and the community survey.
2.4 Relation to other models