Political decisions about what governments collect, publish and use can shape more than statistics. Sebastián León-Giraldo argues that public knowledge is part of democratic infrastructure. When states make some populations harder to see, they can also weaken the evidence used to identify inequality and hold power accountable
Over the last decade, sexual orientation and gender identity have become more visible in some official statistics and administrative systems. Recognition remains recent and uneven, but it has made questions easier to answer. Do trans people, for example, face greater barriers to health care or employment? Are they more exposed to violence?
Recognition, however, can also move backwards. In January 2025, a US executive order instructed federal agencies to use 'sex' rather than 'gender' in official policies and documents. It ordered forms asking for sex to list only ‘male’ or ‘ female’ as options and not provide questions on gender identity; it also directed federal agencies to remove materials promoting what the administration called 'gender ideology'. The decision therefore changed not simply language, but what federal institutions were expected to collect, record and communicate.
Discussions around democratic backsliding usually focus on elections, courts, civil society, the media, and limits on executive power. This series has shown how autocratising governments can target critical gender knowledge. Official data reveal another dimension of this process.
Governments depend on censuses, surveys, health information systems, civil registration and administrative records to understand society. When collected over time and made public, this information creates a shared evidence base. It allows governments to measure differences and track change, while enabling researchers, journalists and citizens to test official claims, and scrutinise public action. The UN Fundamental Principles of Official Statistics hold that official statistics are indispensable to a democratic society's information system.
When official systems cannot identify a population, reliable evidence about its conditions and inequalities becomes harder to produce
This is what I mean by public knowledge: evidence that can circulate beyond the institution that produced it and be used by researchers, journalists, courts, social movements, communities and the public. Governments can use this evidence to design policy, while society can use it to identify inequalities, scrutinise government decisions and demand action. Public knowledge therefore supports public action and democratic accountability. Recognition is one part of this infrastructure: when official systems cannot identify a population, reliable evidence about its conditions and inequalities becomes harder to produce.
Recognition and public knowledge are both reversible. Governments can stop collecting or publishing information, delete data that were available, or fail to use what they already have. Fragmented systems create a similar problem when information held by one institution cannot be connected with other records or translated into public action. What is at stake is more than the loss of a statistical category. It concerns how public institutions produce, preserve, connect, publish and use knowledge.
The democratic value of public knowledge also depends on how that knowledge is produced and governed. For groups exposed to discrimination, data collection can create risks of surveillance, misuse or unwanted exposure. The UN human-rights approach to data therefore emphasises participation, self-identification, disaggregation, transparency, privacy and accountability.
In Colombia, we see the progress and the limits of recognition. DANE, the national statistical office, publishes regular labour-market estimates for the LGBT population using data from the country’s main household and labour-force survey. This makes some socioeconomic inequalities visible.
But recognition remains partial. Colombia’s national LGBTIQ+ policy identifies insufficient inter-institutional coordination as an ongoing problem. There are also reports of serious gaps in information systems that document violence against LGBT people. In my research on mortality data, I show that death certificates do not record gender identity. This has consequences for public policy, institutional recognition and the public record of trans lives. It also makes trans deaths harder to identify in official statistics.
Democratic erosion can affect not only knowledge that already exists, but knowledge institutions were only beginning to make possible
These examples show a fragmented infrastructure of public knowledge: some institutions have started to produce information, while important gaps remain elsewhere. This matters for democratic backsliding because governments do not have to formally reverse recognition to weaken it. Where systems are still being built, political change can stall implementation, stop new data collection or simply leave existing gaps unresolved. Democratic erosion can therefore affect not only knowledge that already exists, but also knowledge that institutions were only beginning to make possible.
This connects with what I describe as statistical justice: the right to be recognised within government systems and for official data to reflect social diversity. Here, the democratic point is narrower: what public institutions are able and willing to know shapes the evidence available for accountability.
Political pressure can affect public knowledge at different points. The aforementioned US example shows how governments can restrict data collection on populations that public institutions had begun to recognise. Brazil showed another route during Covid-19. In June 2020, the Health Ministry removed cumulative case and death totals, along with other detailed information, from its public website. The data returned after a Supreme Court justice ordered full disclosure. Access to public information had itself become a site of institutional conflict.
This is different from revising a measure because evidence shows that it is unreliable. In England and Wales, the Office for National Statistics reclassified the 2021 Census gender-identity estimates as 'official statistics in development' because of concerns about measurement, while continuing research and also publishing guidance on their appropriate use. Statistical systems need this kind of revision to remain credible.
A quality-driven revision of statistical systems seeks to improve what we can know. A politically driven restriction can make populations or problems harder to see
The key question is not whether public information changes, but why it changes, who decides, and through what process. A quality-driven revision seeks to improve what we can know. A politically driven restriction can make populations or problems harder to see. Similar democratic consequences can follow when governments remove information from public access, or fail to connect or use existing data.
This is why public knowledge deserves more attention in debates about democratic backsliding. Democratic erosion can weaken rights and institutions, but also the evidence needed to document its effects in practice. Public data help governments understand society and help society monitor government. When political power begins to decide not only what the state will do, but also which populations the state is allowed to know about, public knowledge itself becomes part of the democratic problem.
