Statistics Finland: use of GSIM

CoSSI model (a national statistical information model)

GSIM Classification model

SIMS

SDMX partly

XBRL (somewhat used)

Statistics Finland is following GSIM development and its implementation in other statistical offices. In 2014, a group of subject matter experts and heads of statistics translated GSIM elements into Finnish. However, this translation was not fully finished as it was not an easy task to find corresponding Finnish terms for all of the GSIM elements. In 2015, some metadata experts and IT architects are taking a closer look at the model, its elements and the counterparts in our daily production.

Use of GSIM Classification model

In 2014, Statistics Finland made the decision to adopt the GSIM Classification model in the new Classification System. See Chapter 2. Business Case for using GSIM

CoSSI: concepts, variables, classifications

GSIM Classification model: classifications

GSIM Classification Model

In 2014, Statistics Finland made the decision to adopt the GSIM Classification model in the new Classification System. Before that, the model was studied and the implementation was designed by a group of classification and IT specialists. International cooperation was also undertaken especially with the Norwegian statistical office. In addition, detailed tests with some individual classifications, as well as comparisons to the national CoSSI model used in dissemination of classifications were made before the decision.

The model fitted, in fact, pretty well to our current system dating back to the beginning of the 1990s, as well as to some newly recognised user needs. Naturally, some amendments were made. The additions to the model were related to the maintenance and usage of classifications. See Chapter 5. Extensions to GSIM. Some of the descriptive classification elements were left out as it was foreseen that the users of the new system would have found these too burdensome. 

Following mappings exist: CoSSI - DDI Codebook (Finnish Social Science Data Archive) and CoSSI - GSIM Classification Model

Folowing mappings under development: GSIM - CoSSI

GSIM Classification model - Extensions in implementation

While implementing GSIM Classification model some amendments were made. The additions to the model were related mainly to the maintenance of classifications including following elements: creator, modifier, validity/status and dates for certain objects. Also some elements describing the usage of classifications and correspondance tables in the statistical business process were added; both the indidual users and the statistics using a classification can be stored in our new system. Some of the descriptive classification elements were left out as it was foreseen that the users of the new system would have found these too burdensome. 

Some special solutions were included in our system. Traditionally, in the Finnish production systems it has been possible to call for a certain classification and its following versions. For this purpose, in addition to classification IDs, we also included a version number indicating a chain of classifications that are time versions of each other. This applies to standard classifications and variants as well.

Another national solution is related to the handling of languages. In Finland, three languages are used in statistical dissemination and data gathering: Finnish, Swedish and English. To support this, the classification item names, as well as names and descriptive data of classification series, classifications and correspondence tables can be easily maintained in all three languages. Also explanatory notes and index entries can be maintained in English and Swedish, in addition to Finnish. In principle, ever more languages can be added into the system.

GSIM

Truly understanding GSIM takes time and requires collaboration with different experts within statistical organizations; subject matter specialists, IT specialists, as well as methodologists are all needed. Nevertheless, this work is worth doing, as GSIM makes cooperation in developing statistical processes smoother and also leads to more applicable outputs of development projects. Common vocabulary, as well as understanding how the information elements are interrelated, is needed both within and between statistical organisations.

GSIM Classification Model

Implementing the GSIM Classification model was a success, which is not a surprise as it is widely used in statistical organisations. The model is pretty well understood among classification experts and IT experts involved in the development of the classification system. The Finnish terminology is also now fixed and this has made communication within our organisation easier; we still need to work on making them widely known among statisticians.

The biggest conceptual challenges lies within the distinction between code lists and classifications. In our system, we treat these pretty equally; others might find that other solutions work better. Another challenge is that the model includes numerous relationships between the elements. Here though, the model is not the one to be blamed but reality itself.

A nationwide challenge still exists in Finland. We have no official national classification model; different governmental organisations use different models and different terms for classifications.