Statistics and Democracy:
A Long Story about Trust and Resentment

The problem with numbers

In January 2022, the New York Times revealed, via a review of internal e-mails, all the ingenious ways that the administration of President Donald Trump had tried to manipulate the results of 2020 United States Census for domestic political gain.[1] Even if most of these plots did not succeed, the story is reminiscent of similar efforts at fudging the numbers on the part of the government of Greece at the start of the 2000s. There the goal had been to misrepresent national debt and deficit statistics so as to sway international politics, specifically that nation’s joining of the European Union. This breach ultimately led the European Union to change the rules on data reporting and to produce new penalties for relaying misleading information,[2] but similar stories, from the cooking of the books on unemployment and poverty numbers in Hungary to the falsification of Covid rates across much of the globe, continue to make news.[3] And conversely, when data reporting is actually accurate—as in the vote tallies after the last US presidential element in 2020—such findings are now often not accepted as such by a wide swathe of the public which has become convinced, with the help of oppositional, right-wing parties, that official information cannot be trusted and measurable victories are hoaxes.[4] Is it any wonder that one of the dominant narratives of our times is that “truth” is singularly in trouble these days, indeed that misinformation, disinformation, and conspiracy thinking have taken over public life?[5]

Rarely, though, have commentators paid much attention to the kind or kinds of information that are heavily contested these days. As in all of these previous examples, disputed information is today often quantifiable and produced by sophisticated, highly trained personnel with expertise in compiling and understanding large data. It is, ironically, the very sort of information traditionally thought to stand for scientific rigor, nonpartisanship, and reliability, especially in contrast to the anecdotal or purely descriptive.

This is not entirely surprising. We are certainly all aware that competing numerical and statistical claims are circulating with new reach, speed, and effect, from social media feeds, to political party missives, to the pronouncements of actual seats of governments. We also know, since it is well-documented by now, that there has been a major breakdown in collective trust in the institutions, methods, and personnel used to generate official statistics and statistical governance. Some of this breakdown in trust is based on legitimate suspicion of partisan tampering, but some is rooted in a growing conviction that all sources of information, no matter the source or means, are politically motivated and thus fail any test of objectivity. Pundits, moreover, have frequently told us in recent years that a fractured data ecosystem, in which people look to sources of information consistent with their preexisting political sentiments and reject others when they do not align, is creating a kind of existential crisis for democracy itself.

And yet, unexpectedly, some basic questions about this situation have gone largely unasked and unaddressed by scholars and policy-makers alike. That includes the question of why so many of our political fights today center on the legitimacy of various statistical findings. Even more, that includes what role statistics and statistical governance play in either challenging our so-called “post-truth condition” or, in fact, fostering it. 

Furthermore, on those few occasions when contextual explanations have been sought, commentators have concentrated almost exclusively on factors that have exacerbated the problem of hostility to all kinds of facts in just the last few years. The standard list includes: a) an explosion of data, much of it generated by unofficial, non-accountable private sources competing with official facts and figures in a saturated marketplace; b) new technological capacities for sharing as well as generating that information which, in the absence of gatekeeping, makes everyone potentially a knowledge creator, distributor, and consumer at once—and with almost immediate, global reach; c) a regulatory and financial arena in which large technology companies that offer platforms for false and often untraceable information, or that use algorithms that boost what is sensational and thus profitable over what is true, are by and large not held liable (and this problem is particularly acute in the US because of the possibility of companies hiding behind an “absolutist” free speech doctrine, though that may be changing)[6]; and d) the fact that all of the above trends have helped fuel political polarization, as people overwhelmed with a superabundance of conflicting and often inaccurate data are most likely to retreat into information subgroups or bubbles characterized by distinctive kinds of epistemic loyalties as well as ideological commitments.[7]  Political leaders, party officials, and emboldened would-be autocrats have worked hard over the past decade to exploit this situation successfully, often with the help of for-profit firms, in wildly divergent parts of the world.  

All of these recent developments are certainly vital to the story of the precarious status of official statistics and, indeed, of knowledge more broadly today. There is also much to say about all of them. But I want to propose that they are inadequate as explanatory factors by themselves. What is needed to fully understand the role of statistics and statistical governance in this particular moment in the history of statistical truth is actually a considerably longer and broader view. To grasp the peculiar situation of quantifiable knowledge creation, dissemination, and utilization today requires looking at history—and a deep history at that, stretching back over centuries rather than mere years. 

Why look so far backward to understand the present, especially when the phenomenon seems so new? First, we need to be reminded that the nature (and not just content) of knowledge, or how individuals come to believe some things to be true and others not, changes in the course of history. It has also always been shaped by and helped shape larger historical currents. The sources, methods, technologies, practitioners, institutions, and even form of truth claims and the conflicts that they generate morph over time in relation to a host of external variables, both structural and contingent, that can be broadly described as social, cultural, political, intellectual, and economic. That is true even for statistical knowledge, which has never actually been able to transcend completely the real world of power relations in which it is crafted, despite what some advocates have always claimed. It simply cannot do so as long as it is the product of particular people with particular social and institutional positions and particular conceptions of the relationship between knowledge and governance (as non-experts have always been aware). Second, the long view matters because the imprint of the past remains in the present. It does so even as communications technology changes, new political actors come into play, and we attach ourselves to the cataclysmic prefix “post” before “truth.” If we ignore long-term patterns in both knowledge production and knowledge use and governance -- patterns that, in some cases, even precede the institution of modern democracy -- we cannot grasp the full story of the twin epistemic and political crises of this present, especially when it comes to the authority of statistics.

What we see if we take this long-term perspective is that statistical knowledge emerged from the start as an alternative to, and potentially in tension with, other forms of knowledge. Those forms included scholasticism and older philosophical traditions, but also contemporaneous folk wisdom and culturally-specific forms of knowledge derived from everyday experience. Moreover, the work of the new cohort of “calculators” associated with early forms of statistical governance almost immediately engendered pushback.  That includes popular skepticism about its legitimacy rooted in an emerging anti-elite epistemic position and a new faith in something called common sense as an antidote.

Some of this conflict was advantageous to the rise of modern representative democracy. Democracies require some squaring off between different sources of truth and different methods for getting there, and when more popular and more elite forms of knowledge production remain in healthy tension but neither is looking to dominate the other, a messy but effective path to reconciliation can result. But since the late eighteenth century, and especially since the end of the nineteenth and the start of the twentieth, expert and non-expert ways of seeing the world have been, long term, on a collision course. Advocates for both forms of truth production have not only steadily expanded their ranks over the last few centuries; they have both become increasingly hostile to the competition, convinced of the illegitimacy of their opponents as truth-tellers and unwilling to look for compromise. The internet has certainly given this process an enormous assist, helping generate hostility and hyper-partisanship in this area as in others. Still, as I hope to show, the basic conflict has been brewing in one form or other since the early modern era. For statistics, as developed by experts, have been bound up with democracy from the start. What follows is my effort to draw attention to this larger and more neglected story of how today’s conflict built from the age of trans-Atlantic democratic revolutions onward.

Deep sources of the current crisis: part I, the story of expertise

Let’s start with the expertise side of the story. Already in the seventeenth and early eighteenth centuries, European rulers, as they conquered new territories and tried to expand their power at home and abroad, began to require new kinds of advisors, with new kinds of knowledge, to make those activities possible. That list included mapmakers, explorers, financial wizards, and military specialists, among others, some positioned inside and some outside of government proper. The growth of secular knowledge production, storage, and use is a vital part of the story of the expansion of modern, centralized states and, especially, imperial states (and though my focus here is primarily on European and European colonial developments, it is not an accident that it is earlier empires--Chinese, Mughal, Ottoman--that developed the most extensive early knowledge bureaucracies).[8] The catchup began in Europe in the early modern era. Already in the sixteenth century, King Philip II of Spain was widely known as “the king of paper” for his prodigious record keeping.[9]

Moreover, the establishment of knowledge bureaucracies in European states meant the emergence of a distinctive and lasting class – a new elite − associated with the provisioning of “useful” knowledge. Even in the earliest modern republics, that is, those formed as the United States and France in the late eighteenth century so deliberately in contrast to the monarchical states that they left behind, it was widely understood from the start that in order to put state-of-the-art knowledge to work, leadership roles had to be assumed by the most virtuous (“men of their word”) and the most wise (“men of knowledge”) and that these were likely one and the same. Thomas Jefferson even spoke, just as the United States was being forged, of the need for a “natural aristocracy.” By that he meant a leadership class associated with truth in both a moral and an epistemic sense to replace one associated with birth. Across the ocean, Mme de Stael promoted the same idea as a social and political palliative in the aftermath of the French Revolution’s Terror.[10]

And from the late eighteenth century onward, it was assumed that much of that useful knowledge would take the form of what was then called in cutting-edge circles “political arithmetic” or “statistics,” literally information useful to states and statesmen.[11] The moment of the birth of the first modern democracies follows closely the moment of the creation of the first national censuses in Denmark, Norway, and Spain. Shortly thereafter, in 1790, the new, independent United States became the first state to constitutionally mandate a regular national census to collect information vital to the nation’s political health. That event was then followed by the introduction of statistics as an academic area of study in German universities and, in the course of the nineteenth century, the advent of national statistical bureaus, which spread from Europe outward, to collect and analyze a growing amount of data about natural and human phenomena alike. Publishing numerical facts about peoples, their environs, and especially their economies came to seem vital to several early democratic promises: making information public and transparent (which early modern state statistics were not); building legislation, including in the economic sphere, upon a foundation of concrete evidence; and apportioning representatives, as well as determining taxation, by accurate population figures.[12]

Thus ironically, just as nation-states in North America, Latin America, and much of Europe were actually democratizing in the nineteenth century in the sense of finally eliminating chattel slavery, extending educational opportunities, and trying out some version of universal manhood suffrage, so did a new kind of knowledge professional, in the wake of this demand for information, appear on the scene. Both on public and, to a lesser degree, private payrolls, men (and I use that word deliberately here) who now called themselves not only “experts” or other derivatives like Fachmann and esperti but also “specialists” and “scientists” and “professionals,” all new coinages of the era, proved themselves more useful than ever.[13] For as governments expanded their purview--in part to compensate for the problems generated by expanding democratization and capitalism--they needed more data, of more kinds, and generated with more sophisticated mathematical tools, to build new kinds of government services, including eventually social welfare policies.[14] At the same time, they also, especially in the case of imperial states like Britain, increasingly saw in quantitative information an aid in efforts to maintain the existing social order and tamp down any hint of rebellion or even dissent.[15] Thus the ranks of those needed to take up the professional mission associated with generating and applying new kinds of reliable, quantitative knowledge grew steadily through the nineteenth century in concert with what Ian Hacking calls “an avalanche of printed numbers.”[16]  And these were just the precursors, it now appears, to the “policy experts” able to make use of this data, like economists, engineers, and urban planners, who would dominate fin-de-siècle and then twentieth-century statecraft domestically, from Japan to Mexico, and in colonial and, eventually, post-colonial realms and transnational bodies alike.[17]

The idea was that impartial, impersonal, quantifiable information, provided by credentialed but nameless experts trained in apolitical, scientific techniques of knowledge gathering, analysis, and dissemination, could be trusted by everyone, including other experts and citizens at large, in an increasingly impersonal world. Their work would serve as an “objective” and often public foundation for productive democratic debate.[18] It would also -- insofar as it marked a departure from information conveyed via more personal and hence idiosyncratic or culturally specific practices of narrative, description, example, and anecdote -- make possible the standardization of results and their exact comparison to the advantage of objective and methodical policy making.  Indeed, at the turn of the new century, the American pragmatist John Dewey imagined the future entailing a perfect blending of democratic and scientific methods and cultures to the advantage of both.[19]

But it was impossible to avoid the anxiety, especially outside centers of statistical governance, that bias and error could still make their way into the data; population numbers, for example, are always vulnerable to political manipulation because the stakes are so high and detection can be exceptionally difficult. Furthermore, there was always a hint, even in the so-called Progressive doctrine of Dewey and colleagues in the late nineteenth and early twentieth centuries, that experts sometimes had to override “the people” for their own good. For one, distrust of the intelligence and knowledge base of ordinary people, and even more of people deemed racially or ethnically inferior, remained prevalent and maybe even grew as what counted as knowledge became simultaneously more specialized, more technical, and more arcane.[20] Moreover, money, it soon became apparent, was always going to matter to the datasphere too, as knowledge has increasingly become a commodity under capitalism, and financial incentives inevitably play some role in shaping its contours. Here we might think of Max Weber’s famous claims from the Weimar period in Germany about democracy and bureaucracy being born together, but also his insistence that the latter would necessarily always prove to be a thorn in the side of the former since they pulled in such opposite directions, not least in terms of their social and economic foundations.[21]

Indeed, the history of the twentieth century now seems rife with examples of states, “developed” and “developing,” that did just that: increasingly move away from popular rule and towards technocracy. Technocracy is simply the name we now give to the modern “planning state” when it seems to be run almost entirely by bureaucrats, specialists, and sometimes businesspeople, all with their own, distinct technical capacities and credentials, a condition that today’s European Union seems, to some observers, to be close to embodying.[22] When Europeans complain about the “democratic deficit” of the EU, they do not mean simply that European “citizens” are not involved in a direct way in making policy since they are not in most places. What they mean is rather that “citizens” have so little influence over the nature of those policies, and so little means to hold officials accountable, and so little even basic understanding of the EU’s technical reports and bills and by whom or how they are crafted, that what gets passed as EU law often seems - as in the frequently evoked example of fishing policies - utterly cut off from the lived experience or everyday “truths” of ordinary Europeans, left or right. That is despite the excellent website, rife with data from gas prices to migration patterns, produced for public consumption by the EU’s own Luxembourg-based statistical agency, Eurostat.[23]

The situation in Washington DC is much the same. Official statistics are generated in voluminous quantities by the Bureau of Economic Analysis, Bureau of Justice Statistics, Bureau of Labor Statistics, Bureau of Transportation Statistics, Economic Research Service, Energy Information Administration, Internal Revenue Service Tax Statistics, National Agricultural Statistics Service, National Center for Education Statistics, National Center for Health Statistics, National Center for Science and Engineering Statistics, Office of Personnel Management, and Office of Research Evaluation and Statistics, not to mention the agencies that monitor all this information, most of it too arcane and specialized for ordinary people to understand even if made publicly available and accessible.[24] The situation is certainly similar in many capitals around the world. And that is only the federal level. States, regions, cites, NGOs, think tanks, academic centers, and private businesses issue yet more such information, often with conflicting numbers. One danger of the trajectory of modern political life is thus that, under the banner of democracy, we have been steadily pushing all truth but technocratic truth, with its close association with statistics, data-driven policy, and the experts who compile and manipulate all of the above, to the side.  The result is, accidentally or not, the erosion of participatory democracy, as well as the trust it requires. It is worth considering whether our dependency on numbers in governing, starting with the establishment of various censuses, can be said to have long been aiding but also undermining democracy simultaneously. 

Deep sources of the current crisis: part 2, Populism

Yet this story about quantification is, as anyone who knows their modern history is aware, only half the story, at best. For resistance both to elites’ dominance of knowledge production and elite ways of knowing, including statistical knowing (with the exception perhaps of sports statistics), also began even before the Age of Revolutions. Pushback came from anti-Enlightenment types who touted not the methods of the scientific revolution applied to the social and political sphere, but rather faith, or instinct, or local traditions, or practical know-how, including what we might now call “the wisdom of the kitchen table,” as less complex and more widely shared ways of getting to truth. Moreover, the common sense of the common people rooted in personal experience, or what the French sometimes called *le bon sens du village*, was an idea that some Enlightenment figures, especially those eager to dissent from various orthodoxies, found useful to harness in their own interest. The idea of common sense made possible both conservative and progressive challenges to established, elite truths using no more than some very basic reasoning and an everyday vocabulary recognizable to all.[25]

Then, in 1776, the Anglo-American revolutionary Thomas Paine, author of the great political pamphlet *Common Sense*, gave this long-simmering alternative epistemology political significance. He announced that, in fact, when it came to the truths of political life, ordinary people not only knew *enough* but actually, in the aggregate, knew better. Then, to make this point, he called on a kind of reverse snobbery in which he suggested that it was precisely the people’s native honesty and plain understanding of things that allowed them to cut through all the phony truisms and absurdities traditionally spouted by church, king, and their fancy, overeducated spokespeople and to get back to basic truths, moral and factual. Moreover, he insisted this everyday common sense provided a foundation on which an entirely new kind of egalitarian, participatory, and radically transparent government could be built.[26] Early efforts in this direction include the first constitution for the state of Pennsylvania and the first Jacobin constitution for France. In both cases, the idea was that the people’s active presence inside government, combined with the elimination of an epistemic ruling class, was essential as protection against inevitable elite abuses of power and as a source of real wisdom.[27]

This idea has survived (though neither radically republican constitution did) down to the present, sometimes associated with the right, sometimes the left. At its best, the repudiation of elite epistemology and the defense of the honesty and wisdom of ordinary people - meaning those outside the realm of educated expertise, which could include women, people of color of both sexes, peasants, immigrants, and/or workers - has served as a way to justify all sorts of emancipatory social and political movements. It has also, as in the case of the civil rights and anti-colonial movements in the 1960s, drawn needed attention to the constant complicity of knowledge and power (though in some ways this was always evident - even the very first census in the United States raised the question, for example, of whether slaves should be counted as people when it came to measuring population and determining representation).[28]

We cannot be surprised either that this kind of skepticism of official truth claims still survives in the general population, as “experts” move frequently between government, academia, NGOs, and political parties, while the broader public understand less and less about the information that these experts provide and the sources, money, or influences behind their claims. All of us are all probably right to maintain some moderate level of skepticism about all information encountered in the public sphere, including what governments say and what they do not not. Covid-19 statistics provide a case in point. The ability to question claims, including official ones, was one of the original reasons for laws protecting freedom of expression and freedom of the press in democracies.[29] And statistics - as Mark Twain made famous by evoking the saying “there are three kinds of lies: lies, damned lies and statistics” - need unpacking as much as any other kind of information if not more, especially since they are generally presented as impervious to politics.[30] Sometimes they even need sabotaging for the sake of social progress, as when many suffragettes famously resisted being counted in the 1911 British Census as long as they were denied the full rights of citizenship, including the vote.[31]

Still, this has not been the only or even dominant form that the challenge to elite or established modes of truth, and especially statistical facts, has taken over the last two hundred plus years, in Britain, in France, in the United States, or anywhere else. Non-expert claims to truth have also often worked quite the opposite way: not to make a claim on behalf of those marginalized by the dominant conception of truth, but instead as a way to reinforce the idea of a single truth position on the part of the “real” or even “true” people, whether of a city, a region, or a nation as whole. For just as claims for elite knowledge formation and representation without the corrective of popular truth can turn exclusionary and lead to the kind of sterile, technocratic government often associated with Brussels or Washington, so arguments for “the people” as the only real source of truth in a democracy can and have run a parallel risk. That risk is increasing disdain for all forms of verifiable, expert knowledge and its purveyors (you might think of British political leader Michael Gove saying, when asked just prior to the Brexit vote to name economists who supported the UK leaving the EU: “I think people have had enough of experts”[32]), as well as for dissenting or outlying voices of all kinds. This is how a society becomes primed for the emergence of a demagogue who perfectly and often cynically incarnates the true will of the “true” people – and potentially chips away at democracy at the same time.

You may also recognize this as a style of politics that now often gets labeled populism whether it is spouted by grassroots protestors or a national leader who has adopted this discourse for self-interested reasons.[33] What needs stressing here is what this discussion has often obscured: that populism depends at its core on a very particular conspiracy story related to democratic truth and its established mode of presentation. The starting point is that some group of people -- often intellectual elites, with or without being in cahoots with an oppressed or marginal group like immigrants -- has usurped the people’s basic, primordial democratic right to define the way the world is. The members of this cohort have done so via obfuscating and impenetrable terms and numbers and phony claims learned at school or abroad.  But, the story goes, when the real or true people finally wake up and realize that everything around them is subterfuge or, today, “fake news,” from the what passes as the scientific conclusions of universities and government agencies to press reports about them, it will be possible to restore the reign of the true people’s everyday common sense, candor, and local knowledge. Then the reign of real-life, consensual solutions to real-life problems will begin at last.

This story too goes back to the very birth of modern democracy; in the 1790s, Edmund Burke was already denouncing the end of the age of chivalry and the success of that of “sophisters, economists, and calculators.”[34] In variants, this narrative has been cropping up sporadically ever since in democracies around the world, from fringe party platforms to presidential rhetoric, on the left and on the right and in between, stoked by a free, commercial press, by domestic and foreign propaganda, and these days, often by specific industries that stand to benefit from capitalizing on it too. Think of all the sites of arguments opposing scientists’ climate change data, from industrial and commerical public relations materials, to political platforms, to on-line civil society groups and how often they turn out to be directly fueling one another both conceptually and financially.[35] Indeed, an anti-elite and anti-expert notion of truth and knowledge seems very recently to have become the fastest growing, if not necessarily yet the most dominant, idiom of politics in the world today, and at a steep price. As problems become ever more complex and transnational, from climate change, to the plethora of migrants and refugees moving around the globe to, most recently, a long-lasting pandemic, the absence of any comprehensible, local solutions, or maybe any viable solutions at all, is driving a kind of nihilism about information that makes well-intentioned publicity for official statistics almost beside the point. We should not neglect either the effects of growing economic dislocation and inequality that have left many people around the world feeling more alienated than ever, culturally and intellectually, from the cosmopolitan but institution-rich world of experts and their ways of knowing. In this context, rejection of even the form of elite knowledge, including statistics, seems all but inevitable. But such epistemic populism does not and cannot exist without its perennial foil, leaving those committed to democratic governance rooted in secure knowledge with a dilemma.

Crafting a response

There are, certainly, some specific measures to combat the assault on legitimate quantitative knowledge, such as that provided regularly by Eurostat, that could prove beneficial. As in the case of the US Census with which this article began, we have ample evidence of how vital it is to protect the independence of statistical agencies, from national censuses to election commissions, from political interference, including by ruling parties or leaders. This is practically ground zero for the defense of democracy and essential to rekindling widespread epistemic trust. But at the same time, such institutions also have an obligation to be more responsive to ordinary citizens than they have been - and that means not just in providing the public (as Eurostat itself has been doing for some time) with simpler, easier-to-interpret data presentations.[36] On the contrary, much of the backlash can be traced to the need expressed by ordinary citizens to be heard and considered and to the failure of ostensibly democratic governments and their enormous bureaucracies to listen. Someone who has just lost his or her job is not heartened to hear that the unemployment rate is down in general; more likely, such persons want to know something specific about their own highly localized prospects, and they want authorities, from government to the media, to take into account how things appear from their on-the-ground vantage point, not just from the top of Mount Olympus. The French Republic famously made ‘expert’ a wide category in rebuilding after the Second World War, including those whose knowledge came largely from practical experience, and gained considerable buy-in across the nation in that way, historians now believe. This was partly the doing of Jean Monnet and his Planning Commission as they sought to combine public participation and ordinary know-how with technical advances.[37] Expertise at the highest levels needs similarly to have broad social contours today.

Then there is schooling and schools, from the most elementary onward, to consider too. One goal for teachers in thinking about civic education in the current age should be to model a healthy, properly democratic skepticism toward established truths of all kinds and to explain why most truths evolve over time. But another aim should be to offer at every level some sense of where and how verifiable knowledge, including statistical knowledge, is produced, where it can be found, how it can be used as well as misused, and of what demonstrability and proof consist. That is another way to start rebuilding public trust in a community of strangers, namely highly-trained and specialized knowledge producers from doctors to career civil servants, and the work they do.[38] A healthy democracy, after all, requires truth--meaning the accurate representation of the world through both words and numbers--as a key collective aspiration.

But at the risk of ending on a thoroughly pessimistic note, we need also to recognize the limitations inherent in any of these plans. Our world has – over the last few centuries – become, simultaneously, so much more democratic in some ways and so much more stratified in others, including economically and educationally, that it is hard to imagine how the longstanding tension between popular and elite modes of knowledge can be overcome. The technological revolution of the internet age has simply intensified both trends and increased the tension between them, now almost to breaking point. As for official statistics: they are not only caught up in the fight but they are an ongoing stimulus of it – and they have been increasingly so as they have come more and more to dominate the official knowledge sphere. That is the great challenge that Eurostat and similar agents of statistical governance confront today. At the very least, seeing this situation clearly, which also means in its historical dimension, should be the starting point for all discussions of remedies. Too bad it rarely is.