The sampling frame used for the TeO2 survey is the 2018 Annual Census Survey (ACS 2018), which includes all people living in municipalities with fewer than 10,000 inhabitants enumerated that year (one in five) and about 8% of the population living in municipalities with more than 10,000 inhabitants.

This is the most recent source available. This makes it possible to have recent addresses, and thus to limit the bias and contact difficulties associated with moves.

In addition, the survey is mainly collected in the census areas that are part of the master sample: this covers almost all urban areas but only part of the rural areas.

Punctual or aperiodic

Computer-assisted collection (Capi) is carried out face-to-face by interviewer.

In addition to the re-translation of some collection documents, the completion of the questionnaire with the help of a third party translator present in the household was authorised and a catch-up survey for TeO2 was set up to allow the completion of the questionnaire with non-French speaking respondents who did not have a third party translator.

Following the health crisis, data collection was interrupted from March to June 2020. It resumed in July 2020, with some interviews conducted by telephone. Telephone interviews were generalised in autumn 2020, following the announcement of the second containment.

Wave 1 collection (immigrants, natives of Drom, general population) took place from July to December 2019.

Wave 2 collection (descendants of immigrants and Dromians) began on 1 January 2020, but was interrupted from 16 March 2020 to 16 June 2020, due to the health crisis linked to the Covid-19 epidemic and the first containment.

Collection resumed in July 2020, with an adapted health protocol, including allowing respondents to complete the interview by telephone in case of vulnerability to Covid-19. On 2 November 2020, following the announcement of the second containment, instructions were given to conduct all interviews by telephone.

From January to August 2020, INED conducted a catch-up survey among non-French speaking individuals who had not been able to be interviewed. For this purpose, INED recruited 9 translators to carry out the survey in 8 languages (Portuguese, Russian, Tamil, Vietnamese, English, Arabic, Chinese and Turkish).

Individual

The main survey sample is composed of the following five sub-samples

• a sample of all individuals aged 18 to 59 living in ordinary households in metropolitan France (general population), i.e. 6,400 respondents ;

• a sample of immigrants (individuals born abroad) aged 18 to 59 living in ordinary households in metropolitan France, i.e. 9,000 respondents ;

• a sample of descendants of immigrants (children with at least one immigrant parent) aged 18 to 59 living in ordinary households in metropolitan France, i.e. 9,000 respondents ;

• a sample of individuals born in an overseas department aged 18 to 59 living in an ordinary household in metropolitan France, i.e. 800 respondents ;

• a sample of individuals with at least one parent in an overseas department, aged 18 to 59, living in ordinary households in metropolitan France, i.e. 800 respondents.

For each of these sub-samples, an optimal respondent size was chosen, and given the expected response rates, a size of the number of people to be surveyed. Among immigrants and descendants of immigrants, it was decided to reconcile an overall representation of all origins with the over-representation of certain less common origins (South-East Asia, Turkey, Central Africa and the Gulf of Guinea, Sahel Africa, etc.). The immigrant sub-sample also includes a group of individuals from countries with large numbers of refugees.

The initial sample consisted of 48,600 address records (AR). More than 2,000 (AR) were added to fill a gap in the initial sample corresponding to individuals born in 1979, 1980, 1989 and 1990 who were absent from the sampling frame. The final allocation of the TeO2 survey was 50,730 AR.

Questionnaire

Information leaflet for respondents

The data are marginally calibrated from the 2019 and 2020 annual census surveys.

The data are adjusted by constructing weights in three steps :

• correction for non-response

• weight sharing

• calibration on margins from the EAR

Not applicable

Not applicable