Data observatories (Fig. 1) were partially triggered by the COVID-19 pandemic. Structures have been built to collect actionable behavioural data from various sources to inform COVID-19-related policy making and communication. These structures continuously collect large-scale data from the public through dialogue formats, social (media) listening methods and surveys6. Ideally, data observatories are set up within national research institutes or agencies at the intersection between science and policy making. They are equipped with a mixed toolbox of methods to collect a wide range of behavioural data. Policy makers and the public shape the research conducted in data observatories. Policy makers highlight knowledge gaps, report challenges and ask questions concerning policy making and communication to the observatories’ scientists. Members of the general public, of specific target groups and of relevant professional groups participate in the various types of studies, panels and formats for dialogue. The observatories study and monitor aspects and determinants of individual and collective action to advance scientific understanding and improve policy making. The behavioural insights gained from this research informs policy making, the implementation of policies and accompanying communication campaigns. In this way, data observatories connect policy makers and the public via behavioural insights.
One task of data observatories is to conduct large longitudinal panel7 or serial cross-sectional studies6 on a regular basis to assess the acceptance of measures, the influencing factors and their potential changes over time. For example, during the COVID-19 pandemic, the World Health Organization (WHO)6,8 recommended regular monitoring of risk perceptions, knowledge, attitudes, self-reported behaviours, levels of trust, psychological strain and misconceptions to identify relevant areas of intervention (for example, when knowledge about transmission modes is low, or when unvaccinated individuals did not trust the vaccines’ safety) and relevant target groups (for example, identifying unvaccinated individuals, or those who are less compliant with health regulations)9. When policy decisions are to be made (for example, regarding a mask mandate or rapid testing regulations), experimental studies or conjoint analyses can be integrated into the surveys10,11 to examine the social and behavioural consequences of such policies.
To help to facilitate climate action, it is crucial to collect this type of situation-specific, large-scale data. Data observatories should assess different facets of the public’s readiness to act — that is, measure public support or rejection of mitigation and adaptation measures, individual willingness to act in a climate-friendly way and willingness to participate in political processes (for example, as done in the Planetary Health Action Survey (PACE)). Notably, by asking people about their concrete current and intended behaviour, their support for climate protection measures and their willingness to politically participate, as well as their reactance towards and current compliance with climate protection measures, PACE goes beyond the usually assessed ‘intentions’ for behavioural change as it targets concrete behaviours and attitudes. Knowing the level of the public’s support for various measures can help policy makers to create bundles of measures that comprise highly accepted and less-accepted measures, along with effective explanations of why change is necessary. To be effective in complementary communication, it is vital to know what affects public support. Although many variables are more or less known from decades of research (for example, social norms, self-efficacy and others), local contexts may vary and render one or the other aspect more important. This is especially important when we consider that over 70% of the research that (psychological) theories build upon is created in a very small number of Western countries. Likewise, it may also be necessary to assess the current distribution of variables of interest in different societal segments (for example, to identify relevant target groups or communities that are most affected by misinformation).
It is crucial that data observatories should first be theory based and theory testing. In this way, generalizable research findings are created and policy advice is built upon evidence as well as theory. Second, the observatories should collect data regularly at short time intervals to monitor changes over time. This enables stakeholders to identify relevant topics and tensions (for example, barriers and enablers of gas saving in the energy crises in Europe). Third, observatories should map the relationship between policy makers and society, for example, how trust in government bodies develops over time or how narratives from campaigns are picked up in public discussions. Fourth, studies conducted in the observatories should be rooted in the current scientific literature of relevant fields, including not only behavioural science such as psychology or economics but also climate science. Fifth, the research questions should be informed by current political and public (media) discussions. These features, in combination, distinguish data observatories from many representative surveys conducted by news outlets or single ministries or governments.
Once such a data collection system is in place, a research team can react quickly to dynamic changes in the situation — for example, when heatwaves are expected, the survey could place special emphasis on heat and assess knowledge gaps and relevant target groups regarding heat protection to improve and tailor health communication efforts. Data collected at various points in time can also be used to detect potential changes in knowledge, self-efficacy or trust (for example, after large education campaigns on television, posters in public spaces or messages on social media). Research designs that use a panel structure offer considerable advantages: within-person changes can be observed, and participants can be selected from certain geographical areas for follow-ups (for example, after floods hit certain areas). On the downside, panels require considerable financial, organizational and administrative resources. However, recent research has shown that using serial cross-sectional designs can also be a good proxy for relative changes over time7.
Other data sources, such as social media or general media data, are also used to generate insights in the aforementioned areas and can fill gaps that cannot be closed by surveys or experiments alone. These allow researchers and governments to track broader narratives, misinformation, commonly discussed topics and widespread sentiments. Internationally, several platforms focus on social media data: a non-exhaustive overview can be found at https://www.parliament.scot/~/media/committ/3435#page=3. Qualitative research approaches are also valuable for examining the breadth of opinions and social consensus. Ideally, data observatories triangulate data from different sources and make them usable for policy advice and health and climate communication.
In its report on behavioural insights and public policy12, the Organisation for Economic Co-operation and Development (OECD) concludes that behavioural insights can indeed improve public policy in many domains, including energy use and environmental protection. The report cites various large-scale online experiments that can improve public policy to mitigate climate change. One could argue that applying the science conducted in the ivory towers of the past few decades may be sufficient to inform policy making and communication activities. Yet, government-funded data observatories would allow for routine use of such tools, complement them with local and current data, and respond quickly to identified knowledge gaps; thus, scaling the effects of evidence-based and theory-informed behavioural public policy making in a timely and topical manner.