12.1 Introduction#
12.1.1 Definition of common statistical infrastructure#
The common statistical infrastructure is defined as the statistical tools and systems that support the activities constituting a generic statistical production process -but that is not part of any specific statistical process- together with the statistical activities required for development or acquisition, maintenance and promotion of these tools and systems. The common statistical infrastructure is part of the so-called intangible assets of a national statistical office (NSO) or any other producer of official statistics within the country.
Examples best illustrate the distinction between activities associated with a generic statistical process and with the common statistical infrastructure:
As the first example, consider the subprocesses of designing and creating the frame for a survey as described in the Generic Statistical Business Process Model (GSBPM) subprocesses 2.4 and 4.1. In the case of a business survey, the frame is typically sourced from a statistical business register (SBR), which is not part of the survey, rather it is developed and maintained for the benefit of all business surveys. Thus, the SBR is an element of the common statistical infrastructure and associated with the statistical activities required to design, develop and maintain it and promote its use. Likewise, an address register is an element of the common statistical infrastructure providing survey frames for household surveys.
As the second example, consider the subprocesses of designing and building a survey questionnaire (GSBPM subprocesses 2.3 and 3.1). They are typically supported by a questionnaire design and construction tool/system independent of the survey. This tool/system is another element of the common statistical infrastructure and is developed or purchased, maintained and promoted, for the benefit of many surveys.
As the third example, consider the subprocess of checking output tables for confidentiality preservation (GSBPM subprocess 6.4). This is typically supported by a confidentiality checking tool, which is also independent of the survey, and which is developed or purchased, maintained and promoted for the benefit of many surveys. This tool is another element of the common statistical infrastructure.
As the fourth example, consider the development, maintenance, and promotion of a statistical standard that defines all the ways a country can be acceptably divided by geographical area to collect and disseminate statistical data. Such a standard is used in the design and dissemination phases of a generic statistical process but is independent of the process. It is another element of the common statistical infrastructure.
12.1.2 Benefits of common statistical infrastructure#
A common statistical infrastructure has two key benefits. First, each element of the infrastructure, once developed, supports all statistical processes, or, at least, all statistical processes of a given type, say business surveys. Thus, a new statistical process can be developed more quickly and at less cost than if there were no common statistical infrastructure, and it can be conducted more efficiently and with greater coherence with other processes. Second, the common statistical infrastructure promotes harmonisation across statistical processes through the use of common methods and standards.
Evidently, the NSO and other large producers of official statistics are likely to have considerably more common statistical infrastructure than smaller producers and, as such, they are likely to derive more benefits. However, smaller producers may also benefit if, as members of the national statistical system (NSS), they can take advantage of common statistical infrastructure developed by the larger members. In fact, the benefits of a common statistical infrastructure apply not only to producers in the NSS, but also to producers in different countries. This is the major aim of UNECE’s Common Statistical Production Architecture (CSPA) (🔗), which is defined as:
“a set of principles for increased interoperability within and between statistical organizations through the sharing of processes and components, to facilitate real collaboration opportunities, international decisions and investments and sharing of designs, knowledge and practices.”
CSPA is further described in Chapter 15.2.12 — Common Statistical Production Architecture (CSPA).
12.1.3 Content and structure of the chapter#
Chapter 7 - Users and their Needs, Chapter 9 - Data Sources, Collection and Processing, Chapter 10 - Analysis and Analytical Frameworks, Chapter 11 - Dissemination of Official Statistics and Chapter 8 - Quality Management describe and provide guidance for the statistical activities associated with a generic statistical production process as described by the GSBPM. The current chapter is complementary to these chapters in the sense that it focuses on the statistical tools and systems that support the activities directly associated with a generic statistical production process while being separate from them. However, it does not cover statistical training, statistical information, data, and metadata management, which are dealt with in Chapter 13 - Human Resources Management and Development and Chapter 14 - Data, Information and Knowledge Management.
A statistical production process starts with a conceptual target population, which is defined in terms of the type of unit that is the subject of the process (for example, enterprise, person, household, farm, etc.) and the envisaged coverage of these units (for example, all enterprises registered for VAT, all persons that are permanent residents of a country or region, etc.).
As discussed in Chapter 9 - Data Sources, Collection and Processing, for a statistical production process that is a survey, the target population is realised in the form of a set of sampling units in a survey frame (abbreviated frame where the context is clear). In principle, the survey frame should provide complete unduplicated coverage of all units included in the target population. In practice, it is the closest approximation to the target population that can be reasonably obtained. The population covered is termed the survey population. The survey frame is not simply a list of sampling units; it includes all the data about those units required for stratification, sampling, and contact.
Survey frames should be coordinated across the surveys that are in any way related so that the data resulting from the surveys are coherent, i.e., can be readily combined without anomalies. For example, the frames for two business surveys intended to cover different industrial sectors should have mutually exclusive coverage.
How a survey frame is constructed depends upon the type of sampling unit involved. In this context, surveys may be divided into four basic types: business surveys, informal sector surveys, agricultural surveys and household surveys.
For most business surveys, the frame is most efficiently and effectively constructed by using a statistical business register (SBR). This describes the construction and maintenance of an SBR and its primary function - that of providing frames for business surveys - and its other uses, particularly as a source of publishable business demographics.
To the extent it depends upon administrative data for construction and maintenance, an SBR cannot provide coverage for informal sector surveys. Frames for such surveys can be obtained from an economic census based on area enumeration, as a by-product of a population and housing census, or using a household-based survey in which a sample of individual households is asked about the businesses.
Although business, informal sector, agricultural and household surveys are treated separately, the three groups are becoming increasingly interrelated.
For example, household surveys may be used to collect data about businesses in the informal sector or identify agricultural producers for own consumption. The SBR may be used as the frame for agricultural surveys of incorporated businesses that are in agriculture. Close cooperation between a population census and agricultural census in a country is strongly advocated in the relevant international guidelines.