How can we understand what a society thinks without relying on traditional surveys that require large samples and significant investment? The Social Probing project, now completed and coordinated by researcher Antonio Fernández Anta at IMDEA Networks, was carried out in collaboration with UC3M, UPM, and Universitat Jaume I to achieve precisely that: develop scalable, cost-effective, and privacy-preserving tools that allow for continuous societal monitoring. Their solution: indirect surveys.

This type of survey doesn’t ask people directly about their own opinions, but about those of their contacts: “How many of your contacts would vote for party X?” or “How many have shown symptoms of COVID-19?” Based on these responses, the team can estimate the actual behavior or situation of the broader population.

Although indirect surveys have existed for decades, their use has been limited. Social Probing, funded by the Ministry of Science, Innovation and Universities, has revived and expanded this methodology, applying it to diverse contexts such as COVID-19 tracking, voting intention estimation in various elections, and the distribution of domestic chores between men and women, in collaboration with the Women’s Institute.

“Thanks to this technique, we’ve been able to accurately estimate voting intentions in regional (such as Madrid or Valencia), national, and European elections. In fact, in the national elections in Madrid, our seat distribution estimate missed just one seat—using only 200 participants—compared to traditional surveys that rely on thousands,” Fernández Anta emphasizes.

Indirect surveys offer several advantages over traditional polls: they require smaller samples, which reduces costs, and they better preserve privacy. “Asking about the voting intention of a participant’s contacts is less intrusive than asking about their own,” explains Fernández Anta. This also increases the likelihood of receiving honest answers.

The team also identified key factors that influence result accuracy, such as the participant’s age, which can introduce hard-to-correct biases. To address this, they have begun stratifying samples by age groups.

The project’s main outcomes include both the data collected and the tools developed to design, deploy, and analyze indirect surveys. This technological and methodological legacy will live on beyond the project: the team is currently working on a PhD thesis focused on robust methods for processing this kind of data, and on identifying which types of social networks are best suited for its application.

They are also exploring new application areas, such as raising awareness about water use in Spain, with new surveys already in development.