While online surveys have distinct advantages, their use in the general population presents several difficulties.
On the one hand, they share the advantages of self-administered questionnaire surveys:

• collection costs are lower essentially because no interviewers are needed,
• respondents can fill the questionnaires whenever it is most convenient for them,
• the absence of interviewers also allows for more personal questions (health, sexuality, etc.).

Online surveys also have their own particular advantages:

• they enable new question formats that integrate video, sound and interactive applications,
• moreover, the collection period can be reduced since there are no (or almost no) limits on the number of people who can be interviewed at the same time. In addition, answers are saved as they are being collected.

However, using the Internet to survey the general population also raises issues about the representativeness of the sample and the extrapolation of results: 

• online surveys are conducted on the basis of volunteer samples, that is, nonprobability surveys,
• people without access to the Internet are effectively excluded. Yet in France in 2012, one out of every five people did not have Internet access at home (Gombault, 2013). 

One way of overcoming these biases is to build an online panel from a random sample of the population. This means recruiting people offline, including people who do not have Internet access, and equipping them with a connection if needed. For the ELIPSS panel this is achieved by using a random sample of addresses drawn by the national institute of statistics and by providing touchscreen tablets with a 3G/4G connection to all the panel members.

Two foreign initiatives inspired the ELIPSS panel project:

• the Longitudinal Internet Studies for the Social Sciences (LISS Panel) of the CentERdata Dutch research institute (Tilburg University)

This online panel representing Dutch households was formed in 2007 on the basis of a probability survey developed in cooperation with the Centraal Bureau voor de Statistiek (CBS, Statistics Netherlands). A simplified computer and Internet connection are provided to households lacking them. The LISS Panel includes 5,000 households, that is, 8,000 individuals aged 16 and over. This system is not free anymore but is still exclusively devoted to research.

• the KnowledgePanel from Gfk (formerly Knowledge Networks) in the United States

Founded in 1999 by two American academics, this system involves random sampling and provision of an Internet connection for respondents who lack one at their home. The sample is randomly drawn from a database of home addresses and involves 50,000 people aged 19 and over. Unlike the LISS Panel, this system is also open to commercial studies.

Since then, similar initiatives were developed through Europe, mainly:

• the German Internet Panel (GIP) at the University of Manheim

This panel represents the population aged 16 to 75 and living in Germany, and is based on a random sampling of more than 4,000 people. Panel members are surveyed every 2 months. A simplified computer or a tablet combined with Internet connection are provided to people who lack them.
This system is reserved for researchers at Manheim University, and the questionnaires focus on political reform.

• the GESIS Panel at the GESIS Leibniz Institute for the Social Sciences in Mannheim

This system uses two types of self-administered questionnaires: 
- people who have home Internet access can answer online,
- people who do not have Internet access or do not want to respond online can respond on paper to questionnaires sent by post. 
The sample is randomly drawn from municipal record and includes 4,900 individuals aged 18 to 70 and living in Germany. The bimonthly surveys are provided by social science research teams through calls for projects, and they cannot serve any commercial interest.

In countries where Internet penetration is very high, above 95% of the population, coverage bias is reduced. That is why the Social Science Research Institute (SSRI) Online Panel established in Iceland in 2010, one part of the Citizen Panel created in Sweden in 2011 and the Norwegian Citizen Panel created in Norway in 2013 chose to recruit panelists from a random sample but not to equip (and therefore do not include) people who do not have internet access at home.

France had no service producing national questionnaire surveys in the humanities and social sciences.

French researchers have two options to fill this gap :

• they can produce surveys by using a survey organization, but this is a costly solution, especially when the survey requires a random sample;

• they can also reuse statistically representative surveys, such as national statistical surveys from the INSEE or from INED, but these do not cover all the subjects and questions that are of interest to researchers.

In addition, survey participation rates have decreased, essentially due to people’s refusals to respond and to the growing difficulties in reaching people to interview. This jeopardizes the statistical quality of surveys.