Case study: Italian National Institute of Statistics - Istat

Istat has adopted a strategy for managing statistical metadata in a harmonized way across the institute. A single centralized system, the Unified Metadata System- SUM, will store and validate all metadata related to statistical data and processes. Furthermore, it integrates quality and metadata management.

The Istat Unified Metadata System is based on the following key concepts and objectives:

-        data retrieval and usability (by associating proper meaning to data as well as, methodological and quality information);

-        metadata reuse (in order to harmonise concepts and reduce documentation burden);

-        traceability (with the objective of statistical process transparency and process automation);

-        integration (with the aim of making the different sectors within a statistical institute speak with one voice and support standardisation)

The SUM system allows for a full description of statistical data and processes and related quality aspects. It is compliant with the international standards such as GSIM and GSBPM as described in the specific case-studies.

It supports the modernization process underway at Istat aimed at overcoming stove-pipe production processes.

The SUM system is composed by two main subsystems:

-        the SIDI-SIQual system, managing metadata describing the statistical production processes and their quality features and quality indicators together with relevant documentation (e.g. methodological manuals, regulations, procedures,…);

-        the SUM-MS system, managing metadata describing all that is necessary to give a definition and a meaning to statistical data (e.g. data content and data structures)

The SIDI-SIQual system was designed back in the late 90s and it is fully implemented. It documents all Istat statistical processes (e.g. surveys, administrative data sources, statistical compilations,..) and manages time series of quality indicators for output quality dimensions for a decade (around 313.000 indicators stored so far).

The SUM-MS system was designed, in its current features, in 2013 and it already contains metadata that define disseminated macrodata for most of the current surveys. In 2015 the system will begin to contain also metadata relative to collected and validated microdata.

The SIDI/SIQual metadata management system is being enhanced in order to be compliant with the ESS conceptual and technical reference metadata standards (SDMX dissemination of metadata according to ESMS and ESQRS).

In Istat approach, metadata are viewed as part of a wider system where different metadata typologies are related to one another and connected to quality.

To this aim, three metadata categories are considered:

-        Metadata related to data structure and content include all that is necessary to give a definition and a meaning to statistical data;

-        Process-related metadata describe the statistical business process in terms of methods (e.g. sampling, collection methods, editing processes) and quality (e.g. timeliness, accuracy).

-        Business-related metadata are useful for the management of an NSI, in planning, executing and assessing both statistical and support activities. These metadata allow stakeholders to connect data and processes with NSI’s strategic objectives (e.g. long-term goals) and with the different management plans of each Institute (e.g. methodological investments); to assign responsibilities and resources to the objectives; to plan schedules and timetables of different actions; to evaluate the achievement of objectives and to assess performance and efficiency.

With regard to SDMX terminology: i) metadata related to data structure and content correspond to respectively structural; ii) conceptual reference metadata and process-related metadata correspond to methodological and quality reference metadata, while iii) business-related metadata are not explicitly considered.

The above mentioned metadata categories find a correspondence to GSIM information objects, as shown in Fig. 1.

Figure 1. Mapping between the GSIM information objects and the proposed classification