How are statistical data collected in France?

-

The Conseil national de l’information statistique (Cnis) is the body that oversees official statistical surveys. By analysing the needs of stakeholders across society, it helps draw up an annual programme of statistical research and surveys aimed at understanding the social and economic dimensions of society. For each proposed new survey, discussions focus on the project's purpose, its place within the information system and the planned arrangements for publishing its findings. The Cnis assesses the relevance of each operation by checking that it meets a need in the public interest and does not duplicate other existing sources.

The Comité du label then ensures compliance with statistical quality criteria. At the same time, any use of sensitive data, even at an intermediate stage, is subject to scrutiny by the Comité du secret statistique. This process is required for every new project and every change to an existing data-processing operation, as well as at regular intervals for projects repeated unchanged.

The population census is based on a sample survey of people living in accommodation, whether in "ordinary" housing or communal establishments. So how can people without accommodation be taken into account? In 2001, 2012 and 2025, Insee and its partners conducted the "Homeless" sample survey, working with associations and support services for homeless people to interview those using their services. For the purposes of the survey, homeless people are those who spent the night before the survey in a shelter or another place not intended for habitation, such as a railway station, metro station, street, park, woodland or car park. Two-stage sampling was used: support services were first selected at random, including accommodation and meal services, day centres and outreach teams, and people attending each selected service on a given day were then sampled at random—for example, by choosing every tenth person in order of arrival. Around 17,000 people responded to the survey in 2025. The first results will be available at the end of 2026.

Insee surveys use home visits and face-to-face interviews, but increasingly also telephone interviews and online questionnaires. This is known as multimode data collection. Its primary benefit is to reduce non-response by reaching different sections of the population as effectively as possible. But each method of data collection also has its own drawbacks, which can produce "biases". In face-to-face interviews, social-desirability bias may arise: respondents give what they believe to be the "right" answer. This method also offers less privacy and therefore carries greater risks when a question is sensitive. In telephone interviews, respondents may make minimal effort, answering quickly and without reflection simply to get the task over with. Online, respondents have no immediate assistance from an interviewer, so they may introduce measurement errors or skip certain questions more readily. No method is perfect!