Open science in democracy research: the research infrastructure “Monitoring Electoral Democracy” (MEDem)

Abstract

Ensuring universal access to scientific research and upholding the principles of keeping data findable, accessible, interoperable, and reusable is of paramount importance to the democratization of science. However, upholding these principles becomes increasingly complex with the increasing scope of data collection, the more different types of data we collect (e.g., survey, text, or institutional and country-level macro data), and the more research teams are involved in data collection. In the domain of democracy research, scientists across Europe are therefore joining forces to launch the research infrastructure monitoring electoral democracy (MEDem), which aims to establish itself as an open platform where the fragmented crowd of researchers in the various research fields can coordinate and develop common standards for data collection both retrospectively as well as prospectively to make their data interoperable, and (comparative) democracy research more productive. Moreover, MEDem will help make democracy research data and findings accessible to the general public (e.g., citizens, journalists, and policymakers).

Introduction

Modern science is fundamentally centered on making research data and findings openly and universally accessible. However, recent advances in digitalization and information technologies have introduced unexpected challenges to this endeavor. While the World Wide Web has significantly facilitated global collaboration and the distributed production of data by researchers across various institutions, the vast proliferation of available findings has made it increasingly difficult to synthesize the larger context. The sheer volume of researchers contributing to knowledge production has grown substantially, along with their collective output. As a result, navigating this extensive and fragmented body of research, and integrating it into a coherent whole to drive meaningful scientific breakthroughs, has become a complex task.

The modern scientific enterprise, therefore, needs new forms of organization to make novel and original discoveries possible. It requires concerted action or an overarching entity that coordinates the distributed data production of disparate researchers and standardizes their data and instruments to make them interoperable. Research Infrastructures (RIs), such as MEDem, can play a critical role in this context. They constitute research facilities that help scattered and individual scientific communities synchronize their efforts to tackle complex and novel research questions together.

Over the past few decades, Research Infrastructures have started to emerge in all parts of the world, covering diverse domains of science. One of the oldest and probably one of the most famous Research Infrastructures is CERN (Conseil Européen pour la Recherche Nucléaire), the laboratory dedicated to understanding what our Universe is made of. Some other well-known infrastructures are, for example and without any claim to completeness, EHT (Event Horizon Telescope), a global initiative to synchronize various telescopes around the world to create a virtual Earth-sized telescope capable of capturing images of black holes; the US-based Open Science Grid, the world’s largest computing grid; the South-African SMCRI (Shallow Marine and Coastal Research Infrastructure), dedicated to collecting research data on South Africa’s coastal zone; AMPC (Asian Mouse Phenotyping Consortium), dedicated to identifying the functions of various genes in the mouse genome; and ALA (Atlas of Living Australia), an infrastructure dedicated to providing free access to biodiversity data, where and when species have been recorded. Particularly relevant to Social Science Research, is, however, the currently evolving Research Data Ecosystem at the Inter-university Consortium for Political and Social Research (ICPSR), which––similar to MEDem––implements a new platform for diverse types of social and behavioral science data to enable novel ways of accessing and analyzing social and behavioral research data.

The European answer to these global developments are the European Research Infrastructure Consortia (ERICs), which are special legal entities under European law that allow for more standardized ways of funding and steering cross-country research efforts. They are collaborative and long-term research facilities that assist in managing research activities that cannot be organized by traditional national or European research funding mechanisms. Within the Life Sciences, they cover, for example, biobanking (BBMRI ERIC), marine biology and ecology research (EMBRC), aerosol, clouds, and trace gases (ACTRIS ERIC), to name only a few. In the Social Sciences and Humanities, as well, several European Research Infrastructures have been established over the last two decades, covering areas such as, for example, data services (CESSDA ERIC), digital language resources (CLARIN ERIC), generations and gender (GGP ERIC), and social surveys (ESS ERIC).

The field of research on democracy and electoral processes, however, is not well-represented, thus far, and is only addressed by existing endeavors in a peripheral way. Yet, today’s democracies face an increasing variety of internal and external challenges, ranging from democratic backsliding and the spread of misinformation to global health, environmental, and military crises. Consequently, the field of Democracy Research produces substantial data that are critically needed to effectively address these challenges. To fully harness their potential and achieve maximum impact, however, improved coordination and increased cooperation among democracy researchers are essential.

In what follows, we will outline the current situation in Democracy Research and present the vision of Monitoring Electoral Democracy (MEDem), the evolving Research Infrastructure that aims at coordinating and harmonizing data (production) in Democracy Research, thus enabling open science and FAIR data. MEDem’s core tasks will bring together and facilitate the linking of various data sources on the functioning of democracies (i.e., survey, text and multimedia, country, and institutional data). In doing so, MEDem aims to promote better, more comprehensive, and highly innovative comparative research on democracies.

A century of collaborative democracy research

The earliest empirical research on democracies already dates back to the nineteenth century. Most notably, however, these studies were case studies or were conducted on only one single country, like, e.g., the pioneering study on “The American Democracy” by Alexis de Tocqueville (1835). The shift toward behavioral studies in the Social Sciences, which gained momentum after World War II, led to an increase in data collections through surveys and experiments.

Among the first multi-country political studies based on observational data was Gabriel Almond and Sydney Verba’s “Civic Culture” (1963), which assessed political attitudes in five nations between 1959 and 1960. Subsequent studies, inspired by the events of the 1968 movements, concentrated on political participation and extended over a limited period only. In due course, however, long-term multi-country studies emerged, addressing broader Social Science topics. Examples include the “Eurobarometer”, which began in 1974 (European Commission 2024), the “European Values Study” (EVS Foundation 2024) and “World Values Study” (World Values Survey Association 2020), commencing in 1981, the “International Social Survey Program”, initiated in 1985 (ISSP Secretariat 2024), and the European Social Survey, which started in 2001 (ESS ERIC and Core Scientific Team 2024b). These studies were characterized by a shift away from reliance on single principal investigators, embracing instead more collaborative governance models. While the Eurobarometer is centrally funded, the other initiatives embody a more decentralized, cross-country infrastructure involving various national research teams, with the ESS being the only one achieving ERIC status thus far.

In the field of election studies, however, systematic collaboration was still rare. Instead, researchers post-harmonized datasets stemming from studies from different countries and different years, leading to insightful but often non-sustainable projects, like, for instance, “The European Voter “ dataset (Andersen et al 2005), which spanned six election studies from 1956 to 1998, the “True European Voter” (Schmitt 2021), which included elections from 29 different countries, or the most recent effort of data harmonization, the “West European Voter” (Garzia et al 2023), which was conducted in the framework of different research projects and included election study data from 16 countries between 1961 and 2020. Yet, post-harmonization poses methodological challenges given variations in questions, scales, and measurements across countries.

Thus, survey researchers attempted to build projects based on a pre-defined questionnaire, starting with the European Election Study (EES) in 1979 (EES Steering Committee 1979-2024). In the 1990s, researchers joined forces to establish the comparative study of electoral systems (CSES) (CSES Secretariat 1996-2024), which incorporated pre-defined questionnaire modules into existing studies with a first module running between 1996 and 2001. The CSES also established procedures of decision-making in rotating planning committees and, pathbreaking at the time, a strict open data policy. Data collection was for the first time not regarded as proprietary to individual researchers but rather as a public scientific good available to all researchers without embargos.

Toward the end of the twentieth century, more and more projects, not exclusively focusing on voting behavior, started to become institutionalized. Examples are the Manifesto Project (MARPOR) (Lehman 2024), which has been studying parties’ policy preferences through election manifestos since 1979, the Comparative Agendas Project (CAP) (Comparative Agendas Project 2015), which started in the early 1990s and developed a standard of coding policy issues in media and government documents, or the Comparative Candidate Survey (CCS) (Lutz et al 2024), which has been conducting candidate surveys since 2005. Increasingly, scholars also started to use expert surveys in political research, like, most notably, the Chapel Hill Expert Survey (CHES) (Jolly et al 2022), which collects data on policy positions of European parties.

Regarding a different aspect, many projects emerged that collected and provided macro-level data in a long-term perspective. For example, the ECPR’s “Political Data Yearbook (PDYi)” (ECPR 2024) started to collect macro-level data, sourced from a network of country experts, on elections and turnout as well as on governments in the early 1990s. Similarly, the Constituency-Level Elections Archive (CLEA) (University of Michigan 2024) archives data from election results worldwide. Currently, many of the larger macro-level datasets with institutional or contextual data at the country/year level are collected in Sweden, such as the Quality of Government (QoG) data, the Varieties of Democracy (V-Dem) project, and Demscore (Demscore 2024), which combines V-Dem and QoG with other Sweden-based data series, like the social policy indicators (SPIN) or the representative democracy data archive (REPDEM).

Toward a more integrated data collection

Many of the projects mentioned above share some common features, although to various degrees: they are collaborative efforts with some structures of governance, they provide data and good documentation in an open way, and they have reached a level of stability over time. However, they also encounter two problems. First, they often lack sustainable and solid funding. Second, the projects have largely not been developed to be interoperable (cf. Mustillo and Springer 2015). They use their own standards of measurement, codes, and documentation, and they cover different years and countries, which limits the usability of data sources in a comparative perspective, or makes their usage time-consuming.

To address these challenges, researchers collaborated in founding ICORE, the “International Committee for Research into Elections and Representative Democracy”, in 1989, which had the mission to enhance cross-national research and foster collaboration among national election study programs. The project was successful in some ways. The aforementioned “The European Voter” dataset was a product of ICORE, and ICORE was also crucial for the establishment of CSES. ICORE, however, did not manage to implement other components or develop into a long-term infrastructure.

This latter aspect was the mission of the PIREDEU project (2008–2011, “Providing an Infrastructure for Research on Electoral Democracy in the European Union”), which was funded through Horizon 2020 and integrated data collection on voters, candidates, the media, and institutions around the 2009 European Parliament elections. PIREDEU did not, however, yet manage to develop into a lasting research infrastructure. The “Consortium for Electoral Research in Europe (CERES)”, which was formed in 2011 out of PIREDEU as a loose network, however, conducted preliminary steps to turn into a European research infrastructure by organizing the community, collecting information on potential project and infrastructure endeavors, among others.

Thus, initiatives like PIREDEU and CERES proved instrumental in shaping MEDem’s path ahead, poised to establish itself as an enduring and indispensable infrastructure for Democracy Research in Europe. MEDem can build upon existing ideas, networks, concepts, and models for an infrastructure and also benefits from the experience of scholars who have been involved in the above-mentioned earlier projects. The first conference toward building the MEDem infrastructure took place in 2017, where many of the key elements were established, like the intention to build an integrated and interoperable infrastructure for different aspects of democracy beyond the study of voting behavior, or the intention to act as a coordinating organization for existing data collection projects. This latter endeavor also led to the creation of the Consortium of National Election Studies (CNES), which was formed in 2022 and aims to promote knowledge exchange between national election studies and to create a permanent harmonized dataset on voter data, or to the funding of OPTED (Observatory for Political Text in European Democracies) that brings together text analysis researchers from different areas of political and communication science. Both, CNES and OPTED are going to be data providing research networks within the MEDem architecture.

Challenges to democracy research

Today’s scholars in Democracy Research have a distinct advantage compared to a few decades ago: digital technologies now allow for collecting large amounts of data at a reasonable cost and sharing them with colleagues around the entire globe. Vice versa, researchers interested in learning about politics, political attitudes, political behavior, political actors, and the like have the means to draw upon the data of others. Many of the existing data are already actively used to inform policymaking and highlight the importance of empirical data to better understand democratic processes. However, making and keeping data FAIR often remains a challenge (cf. Meyers forthcoming), and therefore, the analytical potential of the many data sources is not exploited to its fullest.

Findability of data

Yet, it is difficult to keep track of everything that is available and to identify and find the most suitable data to address one’s particular research interest. This is even more true, as addressing democracies’ most pressing challenges often involves researchers and research data from (at least) two different data sources and sometimes from different disciplines, for example Political Science as well as Communication Science. It requires a lot of experience and familiarity to know who is working on what, or who has collected which data on what topic in which country. There is no central registry or menu one can browse through when looking for existing data or potential collaborators on a particular topic. This lack of findability of data from trusted and high-quality sources unnecessarily wastes the time and resources of researchers who would rather spend them on actual data analysis.

In addition, it creates a hurdle for researchers who newly enter the field and are therefore not yet well acquainted with the existing usances, notably young researchers and researchers from neighboring disciplines who would like to engage in transdisciplinary research projects. If it requires a lot of time and resources to find one’s way in a new field, innovative, novel or even transdisciplinary insights could be hindered.

Accessibility of data and findings

A lot of data is also still not easily accessible because researchers do not share their data or not in a way that gives easy access. In addition, data governance issues as well, mainly data protection and data property rights issues, limit easy access to data. Even more, scientists communicate very often using an insider vocabulary that is not accessible to everyone and often publish their research behind paywalls. Thus, it is many times that it is mainly scientists who have access to scientific findings, while the society at large, journalists, opinion leaders, or policymakers––who could use this information to approach challenges to our democracies––are excluded from the scientific enterprise.

Interoperability of data

Because data collection is scattered across various countries, languages, and numerous independent research teams, and spans several types of data, ranging––at least in the field of Democracy Research––from survey and text data to public statistics, the data at our disposal are often not interoperable. As a result, it often requires a great effort of individual researchers to harmonize data from different countries for comparative analysis, and it requires even greater efforts to link different data objects, like, e.g., surveys and text data, for a joint analysis to understand media effects. Both constitute an obstacle to learning more about the bigger puzzle of social phenomena, tackling more profound and fundamental research questions, and studying them from a comparative perspective across countries and years.

Reusability of data

Many of the data sources as well as the tools and instruments developed to combine and analyze data sets are still not stored in an easily accessible way. This often holds for projects that make data and resources available on their project websites. Once they reach the end of their funding period then, these websites and associated datasets disappear, or they are no longer maintained or updated. Also because of a lack of common standards, datasets and resources are sometimes not sufficiently documented to allow for easy reuse.

Monitoring Electoral Democracy (MEDem)

To come to grips with these challenges, researchers from across Europe are working toward establishing the Research Infrastructure “Monitoring Electoral Democracy”, in short, MEDem. The overall vision of MEDem is to enable more comprehensive and comparative research on electoral democracies by bringing together Democracy Researchers across countries, institutions, and disciplines and enabling the harmonization and linking of their data. In addition, MEDem sets out to open science to the larger public and will dedicate significant efforts to reaching out to citizens, journalists, opinion leaders, and policymakers to supply them with the latest democracy research data and findings in a language and manner that is accessible to people with no prior scientific or statistical training.

MEDem aims to make democracy research data FAIR

As mentioned above, the research community has a wide variety of data at their disposal. MEDem will set out to tame this important, ever-growing amount of democracy research data and make it FAIR to strengthen researchers’ capabilities for exploring novel research problems as well as to enhance their efficiency in data management and data analyses when doing so by saving them from disproportionally large investments of time and resources.

To achieve this, MEDem is, for one, going to host a search facility that allows users to discover all types of data objects––survey data, text and multimedia data, country, and institutional data––that are relevant to their topic (satisfying its mandate for findability and accessibility) and, for another, going to make survey data harmonizable across time and countries and offer linking tools to connect different types of data objects (satisfying its mandate for interoperability and reusability). Regarding data search, MEDem is setting up a searchable knowledge graph that maps the theoretical concepts of all the variables within the MEDem data network (e.g., “partisanship”, or “prospective sociotropic voting”). Users can then search for these theoretical concepts, identify the variables that measure the respective concepts, identify all data sets that contain the variables, and are directed to the repositories that host the desired data sets.¹

In terms of data harmonization, MEDem is developing an automated harmonization service. This service will allow users to select survey datasets for harmonization across countries, years, or both. The MEDem platform will then automatically generate a script for users to run locally, which will guide them through downloading the relevant datasets, recoding variables, handling missing data, applying weighting variables, and merging datasets.² Moreover, MEDem will establish a comprehensive linkage key database.³ This database will contain common linkage keys (e.g., for parties, countries, or media outlets) across different data types, helping users to link the desired datasets.

In particular, the latter aspect of data linking across different data objects (like, e.g., survey data, text data from media coverage or party communication, data about political institutions, or country-level data stemming from official statistics) has the potential to drive Democracy Research to new frontiers and to examine more complex research questions, e.g., on the interplay between public opinion and parties’ or politicians’ behavior, or the way the conventional and new social media mediate these links.

MEDem aims to connect and coordinate Democracy Researchers

Not only is democracy research data fragmented, but so are the communities of democracy researchers. This often impedes their coordination, mutual exchange, and concerted action to tackle larger, more comprehensive, and more complex research questions. MEDem will therefore set out to devise tools to improve communication among researchers and enable them to learn about who is currently working on what, to form expert groups on various topics and share experiences, to early access codebooks and questionnaires, share methodological notes, codes, statistical scripts, info on pretests, and replication data.

MEDem aims to open up Democracy Research to citizens, journalists, and policymakers

MEDem seeks to enhance societal engagement by ensuring responsiveness not only to the needs of researchers but also to the broader public, including citizens, journalists, or policymakers. In this regard, MEDem will offer accessible research data to individuals without prior statistical training. For instance, MEDem will develop a web application that enables users, regardless of their statistical background, to interact with democracy data and create outputs such as tables, cross-tabulations, scatter plots, and line graphs. For more advanced users, such as data journalists and researchers from NGOs or think tanks, MEDem will maintain an information center that will provide guidance on accessing data and contacting principal investigators of specific data collections.

MEDem aims to provide training and capacity building for (early-career) researchers

MEDem will create a dedicated Service Center (SC) for Training and Capacity Building, offering educational opportunities for students at all levels and providing guidance to young scholars on how to make the most of MEDem’s resources in the form of online tutorials, MOOCs, and winter and summer schools to familiarize students and early-career researchers with data linking, data harmonization, and how to analyze large datasets.

MEDem aims to make Democracy Research fit for the future

Technical advancements like, e.g., online surveys, big data analytics, machine learning, artificial intelligence, or quantum computing are about to revolutionize the scientific landscape at an amazing speed (cf. Döring forthcoming). In the years to come, researchers will have to invent new methods, instruments, tools, standards, practices, etc. to keep pace with these developments (cf. Mustillo forthcoming). Major players like research infrastructures, who are equipped with the necessary financial and human capital, might be able to guide and support the field in implementing these technologies and exploit the opportunities they offer for more comprehensive and efficient data collection and analysis.

For example, in the near future, researchers will inevitably need a powerful search facility for identifying the most relevant and trusted data objects in Democracy Research, since it will become impossible to keep track of the enormous volume of data objects that will be available by then. Capitalizing on recent advances in machine learning algorithms, it will be possible to develop a search engine for data objects based on deep neural networks that will help researchers to identify much more precisely than conventional question banks those data objects that are most relevant to their research interest and to provide related-content suggestions. Also, since the user community as well as the data objects are multilingual, the search engine will also need to be trained in neural machine translation to allow users to search for non-English search terms and promote the findability of non-English data objects.

Eventually, MEDem intends to collaborate on this undertaking with researchers from other research infrastructures and neighboring research areas as well, to develop a deep search engine for the larger field of Social Sciences and Humanities more generally.

MEDem’s stakeholder involvement

For MEDem, the next critical step involves establishing its strategic importance within the European Research Area (ERA) by securing a position on the upcoming ESFRI Roadmap 2026. Achieving this objective requires sustained support from diverse stakeholders to ensure MEDem’s long-term financial stability. MEDem is therefore actively seeking political backing from as many countries as possible, with the goal of securing financial contributions through membership fees. At the time of writing, five countries have expressed political support.

As a distributed infrastructure, MEDem also benefits from the financial commitment of prominent research institutions hosting its Service Centers, contributing both infrastructure and expert personnel. At the time of writing, involvement has been shown by an extensive number of research networks. Additionally, MEDem relies on the stable funding of its research networks, which supply the essential data for its operations. At the time of writing, several networks have signed a Memorandum of Understanding, affirming their commitment to collaborate with MEDem. In return, participating networks will benefit from increased data visibility and interoperability, which should enhance data (re)use and strengthen their own funding foundations.

Furthermore, MEDem’s success depends on the broad adoption of its services by the user community. Sustained engagement will require MEDem to closely align its offerings with the needs of both academic and non-academic users. To this end, MEDem hosts workshops, providing a forum for users and stakeholders to discuss their research needs and propose service enhancements. Additionally, MEDem conducted a user survey among nearly 800 researchers who endorsed its establishment (by signing a personal endorsement at the MEDem website), gathering insights on challenges related to data access, harmonization, and linking from multiple sources.

Through these efforts, MEDem seeks to deliver high-impact, user-driven services that reinforce its role as a vital research infrastructure within the ERA and thus to attract sufficient financial support.

Conclusion

The digital revolution offers researchers unprecedented opportunities to collect and analyze vast amounts of data. To fully leverage this potential, a significant restructuring of the currently fragmented scientific enterprise is essential. This involves implementing measures to make large datasets harmonizable and linkable, facilitating collaboration between geographically dispersed researchers and teams. Research Infrastructures, such as MEDem, are well-positioned to provide the sustainable frameworks needed to organize and standardize research activities across disciplines. From a researcher’s perspective, MEDem’s services can streamline the research process by handling preliminary tasks—such as data discovery and integration—that often consume valuable time and resources prior to analysis.

MEDem relies on a stable funding base to ensure long-term sustainability. To achieve this, MEDem aims to establish a diversified funding structure. Firstly, MEDem is supported by several major research institutions, which have committed through a Consortium Agreement to sustain the infrastructure over the long term. MEDem also benefits from the endorsement of various research associations, national infrastructures, and over 800 individual endorsements from researchers worldwide. While the latter support is largely symbolic rather than directly financial, it plays a crucial role in increasing MEDem’s visibility and demonstrating its value to governments, which may, in turn, provide financial support. Additionally, financial contributions from a broad coalition of member countries are essential.

Also, beyond the usual tasks of research projects, Research Infrastructures will have the capacity and resources to provide additional services to the community, like training opportunities for (young) researchers or outreach programs to the non-academic community. In particular, MEDem aims at doing more than just lip service to the inclusion of citizens across audiences and will therefore edit content and research outcome that is understandable and accessible to the general public, ranging from lay people to journalists and policymakers (cf. the ESS Topline Series, ESS ERIC and Core Scientific Team. 2024a). This will foster the resilience of our democracies by fighting misinformation and equipping citizens with scientific findings about their democracies.

Finally, MEDem will serve as an inclusive infrastructure by and for the research community. Its mission to democratize and advance the field of Democracy Research hinges on the active involvement and support of researchers, both as individuals and within their respective country consortia, research groups, or research institutions. This involvement is crucial, since MEDem can only expect solid and sustainable funding as long as it meets the needs of the community. A vibrant user community has the potential to convince funders.