In this new issue: three follow-on papers and a dossier.

Were you intrigued to read “Official statistics and democratic debate: from creation to consolidation (1946–1987)”, in issue N9 last year? Were you waiting for the follow-on edition? Well, here it is, by the same authors. However, one uncertainty remained: when to set the cut-off point for this second installation. Were the authors going to cover a period from 1988 to today? The answer is no: the article by Gaël de Peretti and Béatrice Touchelay ends in the early 2010s, when the influence of digital technology begins to change the game. At the end of the 80s, everything was in place to encourage discussions, to ask questions regarding the uses of statistics and to delve into new themes. The Conseil national de l’information statistique (National Council for Statistical Information, CNIS), a forum that allows and promotes these discussions, replaced the Conseil national de la statistique (National Statistics Council, CNS). However, the context was changing: the European Union was becoming more important and European statistics were imposing multiple constraints on national statistics. Key deficit and debt indicators were playing a structural role. New issues were being addressed, such as homelessness or the search for alternative indicators to GDP. At the same time, different forms of requests for “openness” were emerging: opening up access to individuals’ data for researchers, increased visibility regarding conventions, as well as the provision of all INSEE statistics free of charge, fulfilling the promise of open data ahead of schedule.

Were you surprised by the article on data visualisation, in issue N10, by its story, its methods, its innovations and many examples? The paper by Jocelyne Mauguin and Nicolas Sagnes is a follow-up, since it continues the theme of new methods of dissemination of official statistics. The authors start from a simple observation: INSEE makes an absolutely enormous amount of both statistical data and data on individuals available. Faced with this ocean of data, how can we make life easier for users, so that they can easily access the information they need under good conditions? The answer to this question… depends on the type of users. Cataloguing, simple presentation and data visualisation will take priority for users simply browsing the site and those who use it more intermittently. For experienced users, who want to create their own statistics, the question of formats and, more generally, that of metadata will be essential, as will the possibility of accessing hypercubes that are subject to a high degree of standardisation. The provision of APIs is essential in cases involving a more industrial use of the data, particularly when data are harvested by machines. All this implies changes to the ways in which the site is organised and supplied and the standards it uses: this is the very purpose of the Melodi (Mon espace de livraison des données en open data de l’Insee (My INSEE Space for Receiving Open Data) project.

Are you interested in papers presenting statistics related to an economic sector, such as housing in 2020 (issue N4), or sports or defence in 2023 (issue N10)? The paper by Ronan Le Saout, Nicolas Riedinger and Bérengère Mesqui tackles another specific field, energy. This field naturally lends itself to measurements and quantifications, and this has been the case for centuries. One might think that this makes the task facing Official Statistics considerably easier… but the authors explain that it is not so simple. Conventions must be constantly (re-)defined, on the origin of energy or energy accounting, and nothing ever gets taken for granted. Each energy source has its own specific aspects, which is the case for both production and the link between consumption and price. The statistical observation mechanism must also adapt to new uses, in the context of the ecological and climate transition, such as in the case of the improvement of energy efficiency of housing… as well as to new data sources, such as in the case of smart meters.

The other four articles in this issue of the Courrier form a dossier organised around a single subject: the Répertoire Statistique des Individus et des Logements (Statistical Register of Individuals and Dwellings, RÉSIL) project, which is highly influential for the Official Statistics system’s structure, as it provides the backbone necessary to obtain and perform record linkage between various sources, whether these are administrative in nature or surveys.

In the first article of the dossier, the project owner of the programme, Olivier Lefebvre, presents the RÉSIL project. This paper is of interest in itself, but it also acts as an introduction to the following three articles. Like SIRUS in the world of companies (see issue N8), the register will make it possible to build sample frames or to verify the coverage of administrative data. More generally, its purpose is to produce files that have been enriched by linking records from various sources in a controlled, efficient and responsive manner. To achieve this, RÉSIL relies on several highly diverse pillars: to ensure the quality of the register, to acquire efficient and innovative statistical processing, to benefit from a clear and solid legal basis and, lastly, to obtain a social mandate to be permanently strengthened, that is to say legitimacy, beyond the technical or legal aspects.

The second paper in the dossier, written by Françoise Dupont, Josy Dussart and François Guillaumat-Tailliet, is related to this last point. To ensure the legitimacy of RÉSIL, it has been essential to set up a thorough process of consultation with stakeholders. Generally speaking, this is an essential issue for Official Statistics, yet it receives little focus in the usual literature of the profession. The consultation launched by INSEE, under the aegis of the CNIS, has made it possible to explore and share ideas on what RÉSIL should be and under what conditions it could be used, as well as the ways to explain it to the greatest number of people. It has utilised a range of skills, mostly unrelated to the world of statistics, to take into account ethical considerations (protection of civil liberties, transparency, etc.) and reach a shared assessment of the principles of necessity, minimisation and proportionality, which was more robust and relevant than the initial assessment. INSEE has translated the recommendations of the group involved in the consultation process into legal, technical (in relation to the design and content of the register), organisational and communication provisions. The consultation process, which the article defines in general terms, is not limited to these provisions and the work of communication and listening must continue.

The final two papers concern two techniques that are quite crucial to the RÉSIL resource: record linkage between files and the receipt, acceptance and control of administrative data sources. Both techniques must be used frequently, in a quasi-industrialised manner, and be adapted to various situations, which means that they must be generic, to a certain extent.

The issue of record linkage has already been addressed several times in the Courrier, for example with the système d’information sur l’insertion des jeunes (information system on the integration of young people) (issue N6) or wealth data (issue N7). Here, Heidi Koumarianos, Lucas Malherbe and Olivier Lefebvre examine the subject in a general manner, without specifically focusing on any particular field. The paper can therefore be read both from the perspective of RÉSIL and as an introduction to the issue, with a very broad scope. The authors first note that there are many uses for RÉSIL: to enrich files certainly, but also to lighten the burden of survey questionnaires, to supplement the field of an analysis, to verify the coverage of a source, etc. The article clarifies the vocabulary (matching, interconnection and linkage), describes the legal framework, then presents the methodology, in the event that there is no common identifier between the two files to be matched. It would seem that, even though there is a theoretical framework, a good understanding of the sources remains important in order to configure algorithms effectively. Lastly, to ensure a good statistical control of matched files, the authors stress the importance of measuring the quality of the matching operations.

However, before proceeding with matching, the famous administrative data sources must first be retrieved. This could be considered to be a purely technical matter, a case of merely populating the file, with only questions regarding the format to be settled. In a previous issue (issue N9), the integration of administrative data made it clear that the matter was much broader. What Thomas Tortosa, Manuel Soulier and Olivier Lefebvre describe is a genuine service, allowing the receipt of administrative files within a statistical universe. As a service to be used frequently and in a range of contexts, whether through RÉSIL or not, it must be multifaceted and in particular adaptable, effective, traceable and secure. Data sources are received with a view to pooling them and the receipt phase must be clearly decoupled from subsequent phases, namely the statistical processing operations. The authors place the subject within the more general framework of the General Statistical Business Process Model (GSBPM), to highlight the basic stages to be handled by the accueil-réception-contrôle (receipt, acceptance, control - ARC) tool. The ARC tool was first applied to the déclaration sociale nominative (Nominative Social Declaration, DSN) and it was then applied more broadly, which is essential to the RÉSIL project and raises challenges for other INSEE processes.