The DIME-QUANTI instrument, through its ELIPSS (Longitudinal Online Social Science Survey) panel, was a project that produces questionnaire-based surveys for the scientific community. It was based on a random sample (3000 individual) of the population resident in Metropolitan France, whose members had been supplied with an Internet connected touchscreen tablet on which to answer monthly surveys.
The purpose of Dime Quanti is to give the scientific community the means to produce data to the highest methodological standards.
Because the Elipss sample was randomly drawn, the academic surveys selected was able to achieve the benchmark criteria of statistical quality. In addition, the longitudinal nature of the survey was a strong feature, since there exist very few large-scale panel data for social science research.
A secondary objective of the Dime Quanti project was to develop methodological research, in particular by supporting innovative projects and experiments on survey design.
Finally, the documentation and dissemination of the data produced by the project was intended to promote their reuse and therefore to contribute to cumulative research processes.
Elipss is part of an international landscape of online panels intended for the scientific community. Use of the Internet has developed very quickly in the last 15 years in the field of marketing surveys and opinion polls, mainly because of the low costs and speed of collection possible with the web.
In a report on online surveys (Baker1, 2010), AAPOR (American Association for Public Opinion Research) draws attention to the significant biases in online “access” or “opt-in” panels as a way of conducting general population surveys. Although Internet access has grown rapidly since the early 2000s, part of the population is still off-line. In France in 2017, for example, 15% of the population were still without home Internet access. This lack of coverage automatically excludes part of the population from online surveys. Moreover, the use of access or opt-in panels with non-probabilistic samples casts doubt on their representativeness and on the extrapolation of results.
Nonetheless, the advantages of online surveys very quickly attracted social science research, in particular as a way to compensate for the drop in response rates in traditional surveys.
In order to overcome the biases flagged by AAPOR, an online panel needs to be constructed using a random sample of the population. This means employing an off-line method of recruitment (face-to-face or by telephone) to include people without Internet access and, if necessary, to provide them with an Internet connection (Das, Ester and Kaczmirek, 2011)2.
In consequence, several initiatives emerged in the academic world: the LISS Panel developed by CentERdata at the University of Tilburg in the Netherlands and the Knowledge Panel in the United States were the first to use probabilistic sampling. France with ELIPSS and Germany with GESIS and GIP followed in 2011 (Blom et al., 2016)3. Other probabilistic panels designed for research were then set up in Norway, in Sweden, in Iceland, and in the UK.
The methods used to recruit and include people without Internet access differ between these national schemes. ELIPSS is original in that it gives all the panellists Internet access by supplying them with touchscreen tablets connected to the mobile network. The methodological advantage of this model is that it ensures that all the respondents have the same questionnaire format and that it avoids certain measurement distortions caused by different presentations of the questions.
Submitted in response to calls for proposals, the surveys to be distributed to the ELIPSS panel are selected by the DIME Quanti Scientific and Technical Committee. The ELIPSS team then supports the research teams in the design of the questionnaire and manages the different stages of survey production: scheduling the questionnaire, testing, data collection via the ELIPSS application, field monitoring and reminders, and finally providing the project’s initiators with an exploitable data file.
After a maximum of one year’s exclusive access to the data for the teams that coproduce an ELIPSS survey, the data were added to the CDSP catalogue and were made available to the whole scientific community.
By giving numerous research teams the means to conduct a survey, ELIPSS has collected an impressive array of rich data that sheds light on the most recent changes in social situations, practices and attitudes in France in a very diverse range of fields. Among the 80 resulting datasets, Dime Quanti has given rise to original interdisciplinary collaborations; the SHAMA survey, for example, essentially brought together architects and nutritionists to study the links between housing and obesity. The survey also stood out for the possibility of carrying out longitudinal surveys: the DYNAMOB project with 18 survey rounds conducted on the relationship to politics both during the election period and during the so-called ordinary period, or the annual survey on the evolution of living conditions carried out by the CDSP are the best examples of this. Finally, the data produced within the ELIPSS framework has sometimes been extremely timely, as demonstrated by the project Coping with covid-19.
The richness of these data makes it possible to undertake in-depth analyses of the dynamics running through French society, both because of the very large quantities of individual data collected on the practices, habits and attitudes of the panellists, and because of the longitudinal dimension of some of the surveys, whose questions are repeated several times at regular intervals in order to identify changes in behaviours and opinions at the most granular level.
Beyond the information they provide on French society, the numerous domains explored through the ELIPSS surveys make it possible to explore how this knowledge is constructed and the conditions in which it is produced. In fact, the use of online questionnaires administered to a panel on a touchscreen tablet supplied to the respondents makes it possible both to innovate methodologically, by developing and testing new ways of constructing questions and questionnaires for the general population, and to compare the data produced in this way with those obtained by other survey techniques, in France and in other countries.
The ELIPSS system was based on a variety of areas of expertise (Internet survey design, panel monitoring, protection of respondent confidentiality, development of software infrastructures dedicated to the management of samples, mobile fleets and surveys, etc.) which today must be put to good use in support of new research projects. The experience gained through the implementation and management of the ELIPSS panel can be used for international projects; in fact, since 2019, CDSP teams have been participating in the implementation of the European CRONOS2 panel (led by ERIC ESS) with two projects funded under the H2020 infrastructure calls (SSHOC and ESS-SUSTAIN-2 projects).
In addition, the work undertaken using the ELIPSS experience and data will continue and expand well beyond the initial funding. It will even become clear over time that the value of a longitudinal monitoring system such as ELIPSS is fully justified. In this respect, the device set up to monitor the lockdown experiments during the Covid-19 crisis has shown how valuable it is to have data prior to an event in order to establish what kind of impact it will have.
The sustainability of the ELIPSS system is also guaranteed. Its modalities have changed. In response to changes in the place of the Internet in our society and to limit the overall costs of the system, tablets are no longer being provided to respondents and Internet subscriptions are no longer being offered. ELIPSS now relies on the commitment of Sciences Po and scientific partnerships with institutions wishing to use the platform. For example, the international ISSP survey chose ELIPSS to carry out its French fieldwork in 2020-2021. In 2020, ELIPSS completed the second refresh of its sampling in order to maintain an overall sample size of more than 2,000 participants.
- Baker, R., Blumberg, S. J., Brick, J. M., Couper, M. P., Courtright, M., Dennis, J. M.…Zahs, D. (2010). Research synthesis: AAPOR report on online panels. Public Opinion Quarterly, 74, 711–781 ↩
- Das, M., Ester, P., Kaczmirek, L., (2011), Social and Behavioral Research and the Internet: Advances in Applied Methods and Research Strategies, Routledge. ↩
- Blom A., Bosnjak M., Cornilleau A., Cousteaux A.-S., Das M., Douhou S., Krieger U. (2016), “A Comparison of Four Probability-Based Online and Mixed-Mode Panels in Europe”, Social Science Computer Review, vol.34, n°1, p.8-25 ↩