Eco&Stat n°05-2026 - From an Official Face-to-Face Survey to an Online Probability-based Panel: The Drivers to Opt-in

Online panels are becoming increasingly common, yet recruitment of panellists frequently relies on voluntary opt-in mechanisms that can lead to self-selection bias. This paper proposes a procedure for constructing a probability-based online panel using an official face-to-face survey as the recruitment frame, taking EU-SILC in Luxembourg as a case study. Opt-in behaviour is analysed using insights from Social Exchange Theory and Leverage–Saliency Theory, treating panel participation as a cost–benefit decision driven by respondent engagement, digital access, and other individual characteristics. A rich set of survey variables and paradata is used to model opt-in behaviour. The estimated logit model displays satisfactory predictive performance and highlights the role of engagement-related factors as predictors of opt-in. Panel weights are constructed by combining the inverse of estimated opt-in probabilities with the EU-SILC weights and are shown to improve the representativeness of the resulting panel. Official surveys can support the development of probability-based online panels, offering statistical offices a procedure for digital data collection without compromising data quality.